openwrt/staging/linusw.git
3 years agoramips/mt7621: drop the weak reordering patch
Rui Salvaterra [Tue, 30 Mar 2021 22:59:48 +0000 (23:59 +0100)]
ramips/mt7621: drop the weak reordering patch

In order to fix random hangs on MT7621, we've been selecting
WEAK_REORDERING_BEYOND_LLSC for years [1]. However, these random hangs have been
most likely caused by an oversight in the MIPS implementation of the kernel
memory consistency model, which has already been fixed for some time (and
backported to stable) [2].

Manually refreshed:
321-mt7621-timer.patch

[1] https://git.openwrt.org/?p=openwrt/openwrt.git;a=commitdiff;h=5c971cd6fdd7298a2017bdb6bea870088eddb8b9
[2] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/arch/mips?h=linux-5.4.y&id=42344113ba7a1ed7b5654cd5270af0d5698d8521

Suggested-by: Ilya Lipnitskiy <ilya.lipnitskiy@gmail.com>
Tested-by: Donald Hoskins <grommish@gmail.com>
Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
3 years agofirewall: update to latest git HEAD
Hans Dedecker [Mon, 29 Mar 2021 19:04:06 +0000 (21:04 +0200)]
firewall: update to latest git HEAD

a4355a6 firewall3: clean up the flow table detection logic
edd0dc5 firewall3: create a common helper to find strings in files

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
3 years agoncurses: add screen-256color terminfo
Paul Spooren [Mon, 29 Mar 2021 08:31:21 +0000 (22:31 -1000)]
ncurses: add screen-256color terminfo

The terminfo is required by the popular terminal multiplexer screen and
tmux, offer it by default as the size impact is minimal with 885 Bytes.

Signed-off-by: Paul Spooren <mail@aparcar.org>
3 years agoncurses: split long line of supported terminfo
Paul Spooren [Mon, 29 Mar 2021 08:21:49 +0000 (22:21 -1000)]
ncurses: split long line of supported terminfo

The terminfo files were all in one row which is terrible to read.
Split them over multiple lines to improve readability.

Signed-off-by: Paul Spooren <mail@aparcar.org>
3 years agosunxi: enable wifi module on Pine64 boards
Zoltan HERPAI [Sun, 28 Feb 2021 15:42:05 +0000 (16:42 +0100)]
sunxi: enable wifi module on Pine64 boards

The Pine64(+) and the SoPine64 baseboard has an optional BT+WiFi
module, carrying an RTL8723BS. The wifi part is connected via
SDIO, the BT part is connected via UART. The BT part requires
additional code changes in the drivers so we'll skip that until 5.10.
Aside from a DTS patch to enable the wifi module, add the required
packages into the device definitions.

[    8.996970] RTL8723BS: module init start
[    9.000922] RTL8723BS: rtl8723bs v4.3.5.5_12290.20140916_BTCOEX20140507-4E40
[    9.007965] RTL8723BS: rtl8723bs BT-Coex version = BTCOEX20140507-4E40
[    9.014785] pnetdev = 000000009bdd5000
[    9.047780] RTL8723BS: rtw_ndev_init(wlan0)
[    9.053349] RTL8723BS: module init ret =0

root@OpenWrt:/# iw list
Wiphy phy0
        wiphy index: 0
        max # scan SSIDs: 9
        max scan IEs length: 2304 bytes
        max # sched scan SSIDs: 0
        max # match sets: 0
        Retry short limit: 7
        Retry long limit: 4
        Coverage class: 0 (up to 0m)
        Available Antennas: TX 0 RX 0
        Supported interface modes:
                 * IBSS
                 * managed
                 * AP
                 * monitor
        Band 1:
                Frequencies:
                        * 2412 MHz [1] (20.0 dBm)
                        * 2417 MHz [2] (20.0 dBm)
                        * 2422 MHz [3] (20.0 dBm)
                        * 2427 MHz [4] (20.0 dBm)
                        * 2432 MHz [5] (20.0 dBm)
                        * 2437 MHz [6] (20.0 dBm)
                        * 2442 MHz [7] (20.0 dBm)
                        * 2447 MHz [8] (20.0 dBm)
                        * 2452 MHz [9] (20.0 dBm)
                        * 2457 MHz [10] (20.0 dBm)
                        * 2462 MHz [11] (20.0 dBm)
                        * 2467 MHz [12] (20.0 dBm) (no IR)
                        * 2472 MHz [13] (20.0 dBm) (no IR)
                        * 2484 MHz [14] (20.0 dBm) (no IR)

Signed-off-by: Zoltan HERPAI <wigyori@uid0.hu>
3 years agosunxi: enable MUSB in A53 subtarget
Zoltan HERPAI [Sun, 28 Feb 2021 11:51:27 +0000 (12:51 +0100)]
sunxi: enable MUSB in A53 subtarget

The Pine64(+) and the SoPine64 baseboard has two USB ports. One of
these (the lower) is connected directly to a USB host, the other
one's (upper) PHY is shared with the separate OTG controller. In
order to get it working, MUSB support needs to be enabled.

As there are other targets which compile MUSB support into the kernel
and not package it into modules, the same is done here.

[    1.348760] usb usb5: New USB device found, idVendor=1d6b, idProduct=0002, bcdDevice= 5.04
[    1.357029] usb usb5: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[    1.364250] usb usb5: Product: MUSB HDRC host driver
[    1.369219] usb usb5: Manufacturer: Linux 5.4.99 musb-hcd
[    1.374617] usb usb5: SerialNumber: musb-hdrc.1.auto
[    1.379891] hub 5-0:1.0: USB hub found
[    1.383677] hub 5-0:1.0: 1 port detected
[...]
[  697.299440] usb 1-1: new high-speed USB device number 2 using ehci-platform
[  697.461855] usb 1-1: New USB device found, idVendor=090c, idProduct=1000, bcdDevice=11.00
[  697.470038] usb 1-1: New USB device strings: Mfr=1, Product=2, SerialNumber=3
[  697.477180] usb 1-1: Product: USB Flash Disk

Signed-off-by: Zoltan HERPAI <wigyori@uid0.hu>
3 years agobmips: dts: align LEDs
Álvaro Fernández Rojas [Wed, 24 Mar 2021 10:30:23 +0000 (11:30 +0100)]
bmips: dts: align LEDs

Align all bmips devices LED configurations.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
3 years agobmips: backport accepted pinctrl patches
Álvaro Fernández Rojas [Tue, 2 Mar 2021 19:19:30 +0000 (20:19 +0100)]
bmips: backport accepted pinctrl patches

These patches have been accepted for linux v5.13.
External interrupts not supported for now.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
3 years agobmips: vr-3032u: fix gpio keys
Álvaro Fernández Rojas [Wed, 24 Mar 2021 10:11:47 +0000 (11:11 +0100)]
bmips: vr-3032u: fix gpio keys

These GPIO keys were assigned to incorrect pins.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
3 years agobmips: dgnd3700-v2: fix gpio keys
Álvaro Fernández Rojas [Wed, 24 Mar 2021 10:09:27 +0000 (11:09 +0100)]
bmips: dgnd3700-v2: fix gpio keys

These GPIO keys were assigned to incorrect pins.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
3 years agoucode: update to latest Git HEAD
Jo-Philipp Wich [Mon, 29 Mar 2021 10:50:55 +0000 (12:50 +0200)]
ucode: update to latest Git HEAD

aa9621d compiler: rework switch statement code generation
b5f0de1 vm: add trace mode instruction dump output fixes
0341d64 vm: fix another for-loop memory leak
00d9419 vm: fix further memory leaks in trace mode
20a3763 vm: fix loop variable memory leak in NEXTK/NEXTKV instruction
9a6ef2b lib: prevent use-after-free after uc_shift()
03f1324 object: free prototype object when registering existing ressource type
b3d758b compiler: fix for/break miscompilation
86e3970 lib: fix value refcount of uc_unshift() return value
fe464ea run_tests.sh: allow passing tests to run as arguments
091ae1b compiler: fix another try/catch miscompilation
fcedb19 lib: fix passing uninitialized memory to sigprocmask()
0d0357d vm: fix leaking source context strings in trace mode
7a41fb3 lib: use execvp() in system()

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
3 years agobcm53xx: fix Linksys EA9500 partitions
Rafał Miłecki [Mon, 29 Mar 2021 07:28:47 +0000 (09:28 +0200)]
bcm53xx: fix Linksys EA9500 partitions

Use proper DT binding.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
3 years agokernel: backport 5.13 mtd partitioning changes
Rafał Miłecki [Mon, 29 Mar 2021 06:16:17 +0000 (08:16 +0200)]
kernel: backport 5.13 mtd partitioning changes

1. Use upstream accepted NVMEM patches
2. Minor fix for BCM4908 partitioning
3. Support for Linksys firmware partitions on Northstar

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
3 years agomac80211: merge a few pending tx related fixes
Felix Fietkau [Sun, 28 Mar 2021 10:15:42 +0000 (12:15 +0200)]
mac80211: merge a few pending tx related fixes

Improve performance and fix potential mgmt tx hangs/warnings

Signed-off-by: Felix Fietkau <nbd@nbd.name>
3 years agofirewall3: update to latest git HEAD
Tony Ambardar [Fri, 26 Mar 2021 05:47:36 +0000 (22:47 -0700)]
firewall3: update to latest git HEAD

This includes several improvements and fixes:

  61db17e rules: fix device and chain usage for DSCP/MARK targets
  7b844f4 zone: avoid duplicates in devices list
  c2c72c6 firewall3: remove last remaining sprintf()
  12f6f14 iptables: fix serializing multiple weekdays
  00f27ab firewall3: fix duplicate defaults section detection
  e8f2d8f ipsets: allow blank/commented lines with loadfile
  8c2f9fa fw3: zones: limit zone names to 11 bytes
  78d52a2 options: fix parsing of boolean attributes

Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
3 years agomvebu: Fix mac addresses for GL.iNet GL-MV1000
Sven Eckelmann [Thu, 25 Mar 2021 12:26:15 +0000 (13:26 +0100)]
mvebu: Fix mac addresses for GL.iNet GL-MV1000

The original GL.iNet firmware has two different mac addresses in the
factory/art partition. The first one is for the WAN interface only and the
second one is for both lan0 and lan1.

But the original submission for OpenWrt didn't initialize the mac
addresses of the LAN ports for the DSA device at all. The ethernet mac
address was then used for all DSA ports.

Fixes: 050c24f05c85 ("mvebu: add support for GL.iNet GL-MV1000")
Signed-off-by: Sven Eckelmann <sven@narfation.org>
3 years agoramips: correct switch config of Youku yk1
Shiji Yang [Sun, 21 Mar 2021 09:01:49 +0000 (17:01 +0800)]
ramips: correct switch config of Youku yk1

There are only two lan ports and one wan port on Youku yk1

Fixes: e9baf8265bb8 ("ramips: add support for Youku YK1")
Signed-off-by: Shiji Yang <yangshiji66@qq.com>
[add Fixes:]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agokernel: bump 5.4 to 5.4.108
John Audia [Wed, 24 Mar 2021 14:14:36 +0000 (10:14 -0400)]
kernel: bump 5.4 to 5.4.108

All modifications made by by update_kernel.sh.

Build system: x86_64
Build-tested: ipq806x/R7800
Run-tested: ipq806x/R7800 [only 5.4.107]
No dmesg regressions, everything functional

Run-tested by @timocapa:
 - ramips/mt7621/{Xiaomi R3G, Redmi 2100}
 - Lantiq/XWAY/Fritz!Box 7320/7330

Signed-off-by: John Audia <graysky@archlinux.us>
[squash, edit commit message]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agoath79: fix label_mac for NEC Aterm WG1200CR
INAGAKI Hiroshi [Wed, 24 Mar 2021 15:15:57 +0000 (00:15 +0900)]
ath79: fix label_mac for NEC Aterm WG1200CR

On NEC Aterm WG1200CR, the MAC address for WAN is printed in the label
on the case, not LAN.
This patch fixes this issue.

Fixes: 50fdc0374b ("ath79: provide label MAC address")
Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agomvebu: Add button support for GL.iNet GL-MV1000
Sven Eckelmann [Thu, 25 Mar 2021 12:26:15 +0000 (13:26 +0100)]
mvebu: Add button support for GL.iNet GL-MV1000

The original patch to support this device advertised support for the reset
button and the "switch" in the commit message. But neither were actually
integrated in the device tree or documented anywere.

The button itself is now used to trigger a reset (as described in the
official GL.iNet documentation). The switch itself is registered as BTN_0
like other devices from GL.iNet in ath79.

Fixes: 050c24f05c85 ("mvebu: add support for GL.iNet GL-MV1000")
Signed-off-by: Sven Eckelmann <sven@narfation.org>
3 years agokernel: Move CONFIG_USERIO to generic config
Hauke Mehrtens [Thu, 25 Mar 2021 10:31:23 +0000 (11:31 +0100)]
kernel: Move CONFIG_USERIO to generic config

The CONFIG_USERIO option is unset in multiple target configurations. On
the sunxi target it is activated. Move the kernel configuration option
to the generic kernel configuration.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
3 years agokernel: Deactivate CONFIG_VFIO in generic kernel config
Hauke Mehrtens [Thu, 25 Mar 2021 10:27:31 +0000 (11:27 +0100)]
kernel: Deactivate CONFIG_VFIO in generic kernel config

Instead of deactivating this in every target config, deactivate it once
in the generic kernel config. I was asked for this config option in a
x86 64 build in OpenWrt 21.02.

Fixes: 87046e87e229 ("kernel: add missing kernel config symbol")
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
3 years agokernel: add missing kernel config symbol
Kevin Darbyshire-Bryant [Sat, 27 Mar 2021 08:47:47 +0000 (08:47 +0000)]
kernel: add missing kernel config symbol

Bump to 5.10.26 reversed dependencies on IOMMU for CONFIG_VFIO thus
malta (at least) prompted for this new symbol.

Set it to 'false' in the default config.  rockchip & X86 enable it in
target specific configs.

Thanks to Tony Ambardar for falling over this one

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
3 years agomt76: update to the latest version
Felix Fietkau [Fri, 26 Mar 2021 20:56:26 +0000 (21:56 +0100)]
mt76: update to the latest version

6886b57a1534 mt76: connac: introcuce mt76_sta_cmd_info data structure
e529e8afe22a mt76: mt7921: properly configure rcpi adding a sta to the fw
e4d522776804 mt76: mt7921: fix airtime reporting
be2f67e8d3cb mt76: mt7915: fix key set/delete issue
09a1befde4b7 mt76: fix potential DMA mapping leak
f66f8f41d47b mt76: mt7915: refresh repeater entry MAC address when setting BSSID
035e2f6f1ddf mt76: mt7921: get rid of mt7921_mac_wtbl_lmac_addr
ee29cd5f3a6a mt76: mt7615: only enable DFS test knobs for mt7615
9a98b1a6f9c2 mt76: mt7615: cleanup mcu tx queue in mt7615_dma_reset()
3bd285424e7b mt76: mt7622: trigger hif interrupt for system reset
bf6d9ee4acd1 mt76: mt7615: keep mcu_add_bss_info enabled till interface removal
115b74282314 mt76: mt7915: keep mcu_add_bss_info enabled till interface removal
57432e701d1a mt76: mt7915: cleanup mcu tx queue in mt7915_dma_reset()
a519c49a6a42 mt76: mt7615: 0-terminate firmware log messages
4a22f2ffae2e mt76: mt7915: 0-terminate firmware log messages
b8609066893a mt76: mt7615: fix chip reset on MT7622 and MT7663e

Signed-off-by: Felix Fietkau <nbd@nbd.name>
3 years agox86/base-files: add support for Sophos SG/XG-105
Stan Grishin [Wed, 24 Mar 2021 16:18:08 +0000 (16:18 +0000)]
x86/base-files: add support for Sophos SG/XG-105

This adds detection of the Sophos SG-105 and Sophos XG-105 models
and assignment of ethernet ports these models have to LAN/WAN.

Signed-off-by: Stan Grishin <stangri@melmac.net>
3 years agoopenssl: bump to 1.1.1k
Eneas U de Queiroz [Fri, 26 Mar 2021 17:46:29 +0000 (14:46 -0300)]
openssl: bump to 1.1.1k

This version fixes 2 security vulnerabilities, among other changes:

 - CVE-2021-3450: problem with verifying a certificate chain when using
   the X509_V_FLAG_X509_STRICT flag.

 - CVE-2021-3449: OpenSSL TLS server may crash if sent a maliciously
   crafted renegotiation ClientHello message from a client.

Signed-off-by: Eneas U de Queiroz <cotequeiroz@gmail.com>
3 years agoREADME: add install command to build requirements
Petr Štetiar [Thu, 25 Mar 2021 07:01:08 +0000 (08:01 +0100)]
README: add install command to build requirements

In commit 8c8496435a9d ("build: add GNU install to prerequisites") new
dependency was added, but addition to README was omitted.

Fixes: 8c8496435a9d ("build: add GNU install to prerequisites")
Signed-off-by: Petr Štetiar <ynezz@true.cz>
3 years agoRevert "openssl: refresh patches"
Kevin Darbyshire-Bryant [Fri, 26 Mar 2021 09:12:12 +0000 (09:12 +0000)]
Revert "openssl: refresh patches"

This reverts commit e27ef2da0d513494c3e9926ce8d44b63e4236a32.

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
3 years agoopenssl: refresh patches
Kevin Darbyshire-Bryant [Fri, 26 Mar 2021 08:58:07 +0000 (08:58 +0000)]
openssl: refresh patches

Tidy up some patch fuzz.

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
3 years agoelfutils: enable building with MIPS16
Tony Ambardar [Wed, 24 Feb 2021 16:52:22 +0000 (08:52 -0800)]
elfutils: enable building with MIPS16

Building with MIPS16 was disabled in 2013 due to an issue with GCC TLS:
https://dev.archive.openwrt.org/ticket/13572. But after the problematic
GCC version was retired, this change wasn't revisited.

Re-enable MIPS16 builds to reduce average elfutils library sizes ~10%.
This was compile-tested on malta/mips32be and malta/mips32le, and linked
with iproute2 for run-testing. Package sizes follow:

Library  MIPS16:=0  MIPS16:=1
-------  ---------  ---------
libelf1    43217      37492
libasm1    12481      11658
libdw1    229723     205793

Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
3 years agokernel: bump 5.10 to 5.10.26
Rui Salvaterra [Thu, 25 Mar 2021 09:32:18 +0000 (09:32 +0000)]
kernel: bump 5.10 to 5.10.26

Automatically refreshed:
ath79/patches-5.10/471-mtd-cfi_cmdset_0002-AMD-chip-0x2201-write-words.patch
bcm63xx/patches-5.10/143-gpio-fix-device-tree-gpio-hogs-on-dual-role-gpio-pin.patch
generic/hack-5.10/204-module_strip.patch

Run-tested:
ath79 (TL-WDR3600)
x86_64 (APU2 - ldir)

Tested-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
3 years agobuild,json: 3rd fixup of default_packages
Paul Spooren [Thu, 25 Mar 2021 20:25:54 +0000 (10:25 -1000)]
build,json: 3rd fixup of default_packages

This became a bit of a tragedy, caused by a corner cases which wasn't
put into account during testing. DEFAULT_PACKAGES are defined in
target/linux/<target>/Makefile but also in
target/linux/<target>/<subtarget>/target.mk.

The latter was no longer imported when using DUMP=1, however not using
DUMP=1 while running the Makefile in target/linux/<target>/ caused duplicate
packages in the list.

As a solution, which should have been used from day 0, `make` runs in
target/linux/ without DUMP=1, resulting in no duplicate packages and all
inclusions from include/target.mk, linux/target/<target>/{Makefile,
<subtarget>/target.mk}

While at it, sort the list of default packages.

Signed-off-by: Paul Spooren <mail@aparcar.org>
3 years agobuild,json: fixup fixup of arch_packages
Paul Spooren [Thu, 25 Mar 2021 04:54:36 +0000 (18:54 -1000)]
build,json: fixup fixup of arch_packages

The commit "1bf2b3fe90 build,json: fixup missing arch_packages" fixes
the missing package architecture locally but runs $(TOPDIR)/Makefile
rather than a target specific one. While this works on local builds just
fine, it causes the buildbots to add garbage to the `arch_packages`
variable:

    cd \"/builder/shared-workdir/build\"; git log --format=%h -1
    toolchain > /builder/shared-workdir/build/tmp/.ver_check\ncmp -s
    /builder/shared-workdir/build/tmp/.ver_check
    /builder/shared-workdir/build/staging_dir/toolchain-x86_64_gcc-8.4.0_musl/stamp/.ver_check
    || { \\\n\trm -rf
    /builder/shared-workdir/build/build_dir/target-x86_64_musl
    /builder/shared-workdir/build/staging_dir/target-x86_64_musl
    /builder/shared-workdir/build/staging_dir/toolchain-x86_64_gcc-8.4.0_musl
    /builder/shared-workdir/build/build_dir/toolchain-x86_64_gcc-8.4.0_musl;
    \\\n\tmkdir -p
    /builder/shared-workdir/build/staging_dir/toolchain-x86_64_gcc-8.4.0_musl/stamp;
    \\\n\tmv /builder/shared-workdir/build/tmp/.ver_check
    /builder/shared-workdir/build/staging_dir/toolchain-x86_64_gcc-8.4.0_musl/stamp/.ver_check;
    \\\n}\nx86_64

Only the last line contains the desired string.

Future investigation should check why the build system prints this to
stdout rather than stderr.

Signed-off-by: Paul Spooren <mail@aparcar.org>
3 years agobuild: add GNU install to prerequisites
Kevin Darbyshire-Bryant [Mon, 4 Jan 2021 20:55:38 +0000 (20:55 +0000)]
build: add GNU install to prerequisites

The recent removal of usbutils from core and replacement by hwdata in
packages has exposed hwdata's requirement for certain GNU options on
'install' (-T)  Other packages (sqm-scripts) have openwrt specific
makefile sections to avoid GNU options but I suspect this is going to
get harder in the future.

Add GNU install as a prerequisite and link into
$STAGING_DIR/host/etc/bin as per similar GNU utils

This resolves an issue building under MacOS which would otherwise use a
non-GNU options aware version of 'install'

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
3 years agobuild: make sure asm gets built with -DPIC
Philip Prindeville [Fri, 19 Mar 2021 21:09:45 +0000 (15:09 -0600)]
build: make sure asm gets built with -DPIC

Fixes issue openwrt/packages#14921, whereby inline ASM wasn't getting
built as PIC; look at gmp-6.2.1/mpn/x86/pentium/popcount.asm for
example:

ifdef(`PIC',`
...

for a routine that exists in both PIC and non-PIC versions.

Make sure that wherever $(FPIC) gets passed as a variable expansion
that it gets quoted where necessary (such as setting environment
variables in shell commands).

Signed-off-by: Philip Prindeville <philipp@redfish-solutions.com>
3 years agobuild,json: fixup missing arch_packages
Paul Spooren [Wed, 24 Mar 2021 21:01:46 +0000 (11:01 -1000)]
build,json: fixup missing arch_packages

Fix 7f4c2b1 "build,json: fix duplicates in default_packages" which
removed duplicate default packages but also removed the package
architecture from the profiles.json.

If DUMP=1 is set, the `ARCH_PACKAGES` is no longer exported and
therefore empty. Fix this by running make twice, once with DUMP=1 and
once without.

Signed-off-by: Paul Spooren <mail@aparcar.org>
3 years agoimage: fix append-image when building multiple profiles
Daniel Golle [Wed, 24 Mar 2021 22:53:23 +0000 (22:53 +0000)]
image: fix append-image when building multiple profiles

In case CONFIG_TARGET_MULTI_PROFILE is set, IMG_PREFIX cannot be
expanded. Use DEVICE_IMG_PREFIX instead and make sure it's defined.

Fixes: 8f89b1ab0f ("image: add 'append-image' build command")
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
3 years agofirmware-utils: bcm4908kernel: name struct fields
Rafał Miłecki [Wed, 24 Mar 2021 21:44:14 +0000 (22:44 +0100)]
firmware-utils: bcm4908kernel: name struct fields

Less magic names / values.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
3 years agokernel: create bootfs partition when parsing on BCM4908
Rafał Miłecki [Wed, 24 Mar 2021 17:20:13 +0000 (18:20 +0100)]
kernel: create bootfs partition when parsing on BCM4908

It's helpful for accessing booting data (DTS, kernel, etc.). It has to
be used carefully as CFE's JFFS2 support is quite dumb. It doesn't
recognize deleted files and has problems handling 0 inode.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
3 years agofirmware-utils: bcm4908img: name fields & values
Rafał Miłecki [Wed, 24 Mar 2021 16:29:30 +0000 (17:29 +0100)]
firmware-utils: bcm4908img: name fields & values

Less magic numbers

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
3 years agomediatek: generate complete sdcard image for BPi-R64
Daniel Golle [Wed, 24 Mar 2021 13:44:52 +0000 (13:44 +0000)]
mediatek: generate complete sdcard image for BPi-R64

Populate the recovery and production partitions of the generated sdcard
image for the Bananapi BPi-R64.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
3 years agoimage: add 'append-image' build command
Daniel Golle [Wed, 24 Mar 2021 13:34:13 +0000 (13:34 +0000)]
image: add 'append-image' build command

Commit 7ce1d9ce09 ("build: artifacts add dependency for built images")
now makes sure that sysupgrade and initramfs images are available at
the stage that artifacts are created.
Allow making use of that with a new build command 'append-image' to
be used in artifacts.
See the next commit for an example.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
3 years agobuild: artifacts: add dependency for built images
Oskari Lemmela [Sun, 21 Mar 2021 08:58:10 +0000 (10:58 +0200)]
build: artifacts: add dependency for built images

Add possibility to use images and initramfs in artifacts.

Signed-off-by: Oskari Lemmela <oskari@lemmela.net>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
3 years agomwlwifi: add PKG_FLAGS:=nonshared
Daniel Golle [Wed, 24 Mar 2021 09:47:12 +0000 (09:47 +0000)]
mwlwifi: add PKG_FLAGS:=nonshared

This should fix the problem of mwlwifi-firmware-* not being found
when using the ImageBuilder.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
3 years agoucode: fix PKG_MIRROR_HASH
Daniel Golle [Wed, 24 Mar 2021 14:13:43 +0000 (14:13 +0000)]
ucode: fix PKG_MIRROR_HASH

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
3 years agobmips: reorganize patches
Álvaro Fernández Rojas [Wed, 24 Mar 2021 13:49:46 +0000 (14:49 +0100)]
bmips: reorganize patches

Reorder kernel patches after recent backports.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
3 years agobmips: switch to upstream ehci overcurrent flag
Álvaro Fernández Rojas [Wed, 24 Mar 2021 13:19:12 +0000 (14:19 +0100)]
bmips: switch to upstream ehci overcurrent flag

After backporting upstream ehci overcurrent patches we need to use spurious-oc
instead of ignore-oc.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
3 years agokernel: backport ehci overcurrent patches
Álvaro Fernández Rojas [Wed, 24 Mar 2021 13:02:56 +0000 (14:02 +0100)]
kernel: backport ehci overcurrent patches

These patches have been accepted for v5.13.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
3 years agokernel: drop apu2 reboot patch
Kevin Darbyshire-Bryant [Tue, 14 Jan 2020 22:19:06 +0000 (22:19 +0000)]
kernel: drop apu2 reboot patch

This patch really annoys me, either it needs to go upstream or be
dropped, so it's going to be dropped here.

Checking drivers/platform/x86/pcengines-apuv2.c it also appears to be
incomplete since it mentions different dmi board names depending on bios
version.

/* APU2 w/ legacy BIOS < 4.0.8 */ is 'APU2'
/* APU2 w/ legacy BIOS >= 4.0.8 */ is 'apu2'
/* APU2 w/ mainline BIOS */ is 'PC Engines apu2'

So the patch, if applicable at all, only 'works' for legacy BIOS >=
4.0.8

My APU2 on mainline BIOS reboots fine without this patch.  So let's see
if anyone screams and when they do question why legacy bios.  If patch
DOES need to be re-introduced then it needs to go upstream first.

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
3 years agokernel: bump 5.10 to 5.10.25
Kevin Darbyshire-Bryant [Sat, 20 Mar 2021 11:44:57 +0000 (11:44 +0000)]
kernel: bump 5.10 to 5.10.25

Straightforward refresh of patches using update_kernel.

Removed (reverse-applicable):
bmips/patches-5.10/010-v5.11-net-dsa-implement-a-central-TX-reallocation-procedur.patch

Run tested: x86_64 (apu2)

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
3 years agokernel: bump 5.10 to 5.10.24
Kevin Darbyshire-Bryant [Fri, 19 Mar 2021 15:21:25 +0000 (15:21 +0000)]
kernel: bump 5.10 to 5.10.24

Straightforward refresh of patches using update_kernel.

Run tested: x86_64 (apu2)

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
3 years agobmips: add wifi packages for supported devices
Álvaro Fernández Rojas [Tue, 23 Mar 2021 19:23:47 +0000 (20:23 +0100)]
bmips: add wifi packages for supported devices

Netgear DGND3700v2 / Comtrend VR-3032u internal wireless not supported.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
3 years agobmips: add ATH9K PCI fixups
Álvaro Fernández Rojas [Tue, 23 Mar 2021 19:16:37 +0000 (20:16 +0100)]
bmips: add ATH9K PCI fixups

Add support for registering ATH9K PCI fixups needed to bring up wifi on some
devices.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
3 years agobmips: add B43 SPROM PCI fixups
Álvaro Fernández Rojas [Tue, 23 Mar 2021 19:14:38 +0000 (20:14 +0100)]
bmips: add B43 SPROM PCI fixups

Add support for registering fallback BCMA/SSB B43 SPROMs.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
3 years agobmips: add experimental PCI/PCIe support
Álvaro Fernández Rojas [Tue, 23 Mar 2021 19:12:22 +0000 (20:12 +0100)]
bmips: add experimental PCI/PCIe support

These PCI drivers are a bit hacky and definitely not suitable for upstreaming,
but hopefully we can use them as a base for developing proper upstream PCI
drivers.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
3 years agorealtek: add ZYXEL_VERS to DEVICE_VARS
Adrian Schmutzler [Mon, 22 Mar 2021 19:58:16 +0000 (20:58 +0100)]
realtek: add ZYXEL_VERS to DEVICE_VARS

Otherwise, the last defined value will be set for all devices.

Fixes: c6c8d597e183 ("realtek: Add generic zyxel_gs1900 image definition")
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agoramips: speed up spi frequency for Youku YK-L1
Shiji Yang [Sun, 21 Mar 2021 14:43:59 +0000 (22:43 +0800)]
ramips: speed up spi frequency for Youku YK-L1

Youku YK-L1 has a huge storage space up to 32 MB. It is better to
use a higher spi clock to read or write serial nor flash chips.
Youku YK-L1 has Winbond w25q256fvfg on board that can support
104 MHz spi clock so 48 MHz is safe enough.
The real frequency can only be sysclk(580MHz ) /3 /(2^n) so 80 MHz
defined in dts file will set only 48 MHz in spi bus.

Signed-off-by: Shiji Yang <yangshiji66@qq.com>
3 years agoath79: copy cfi patch from 5.4 over to 5.10
Mauri Sandberg [Mon, 22 Mar 2021 14:36:58 +0000 (16:36 +0200)]
ath79: copy cfi patch from 5.4 over to 5.10

Add the same patch to 5.10 too. The patch is in process of being
upstreamed.

Fixes: 8cc0fa8facea ("ath79: cfi: cmdset_0002: amd chip
0x2201 - write words")

Signed-off-by: Mauri Sandberg <sandberg@mailfence.com>
[add Fixes:]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agoath79: Add support for Buffalo WZR-HP-G300NH
Mauri Sandberg [Mon, 7 Sep 2020 19:04:26 +0000 (22:04 +0300)]
ath79: Add support for Buffalo WZR-HP-G300NH

This device is a wireless router working on 2.4GHz band based on
Qualcom/Atheros AR9132 rev 2 SoC and is accompanied by Atheros AR9103
wireless chip and Realtek RTL8366RB/S switches. Due to two different
switches being used also two different devices are provided.

  Specification:
  - 400 MHz CPU
  - 64 MB of RAM
  - 32 MB of FLASH (NOR)
  - 3x3:2 2.4 GHz 802.11bgn
  - 5x 10/100/1000 Mbps Ethernet
  - 4x LED, 3x button, On/Off slider, Auto/On/Off slider
  - 1x USB 2.0
  - bare UART header place on PCB

  Flash instruction:
  - NOTE: Pay attention to the switch variant and choose the image to
    flash accordingly. (dmesg / kernel logs can tell it)
  - Methods for flashing
    - Apply factory image in OEM firmware web-gui.
    - Sysupgrade on top of existing OpenWRT image
    - U-Boot TFPT recovery for both stock or OpenWRT images:
      The device U-boot contains a TFTP server that by default has
      an address 192.168.11.1 (MAC 02:AA:BB:CC:DD:1A). During the boot
      there is a time window, during which the device allows an image to
      be uploaded from a client with address 192.168.11.2. The image will
      be written on flash automatically.

      1) Have a computer with static IP address 192.168.11.2 and the
         router device switched off.
      2) Connect the LAN port next to the WAN port in the device and the
         computer using a network switch.
      3) Assign IP 192.168.11.1 the MAC address 02:AA:BB:CC:DD:1A
         arp -s 192.168.11.1 02:AA:BB:CC:DD:1A
      4) Initiate an upload using TFTP image variant
         curl -T <imagename> tftp://192.168.11.1
      5) Switch on the device. The image will be uploaded subsequently.
         You can keep an eye on the diag light on the device, it should
         keep on blinking for a while indicating the writing of the image.

  General notes:
  - In the stock firmware the MAC address is the same among all
    interfaces so it is left here that way too.

  Recovery:
  - TFTP method
  - U-boot serial console

  Differences to ar71xx platform
  - This device is split in two different targets now due to hardware
    being a bit different under the hood. Dynamic solution within the same
    image is left for later time.
  - GPIOs for a sliding On/Off switch, marked 'Movie engine' on the device
    cover, were the wrong way around and were renamed qos_on -> movie_off,
    qos_off -> movie_on. Associated key codes remained the same they were.

  The device tree source code is mostly based on musashino's work

Signed-off-by: Mauri Sandberg <sandberg@mailfence.com>
3 years agoath79: cfi: cmdset_0002: amd chip 0x2201 - write words
Mauri Sandberg [Mon, 14 Sep 2020 19:45:15 +0000 (22:45 +0300)]
ath79: cfi: cmdset_0002: amd chip 0x2201 - write words

Generally, in upstream CFI flash memory driver uses buffers for write
operations. That does not work with AMD chip with id 0x2201 and we must
resort to writing word sized chunks only. That is, to not apply general
buffer write functionality for this given chip.

Without the patch kernel logs will be flooded with entries like below:

MTD do_erase_oneblock(): ERASE 0x01fa0000
MTD do_write_buffer(): WRITE 0x01fa0000(0x00001985)
MTD do_erase_oneblock(): ERASE 0x01f80000
MTD do_write_buffer(): WRITE 0x01f80000(0x00001985)
MTD do_write_buffer_wait(): software timeout, address:0x01f8000a.
jffs2: Write clean marker to block at 0x01a60000 failed: -5
MTD do_erase_oneblock(): ERASE 0x01f60000
MTD do_write_buffer(): WRITE 0x01f60000(0x00001985)
MTD do_write_buffer_wait(): software timeout, address:0x01f6000a.
jffs2: Write clean marker to block at 0x01a40000 failed: -5

References: http://patchwork.ozlabs.org/project/linux-mtd/patch/20210309174859.362060-1-sandberg@mailfence.com/
Signed-off-by: Mauri Sandberg <sandberg@mailfence.com>
[added link to usptream fix submission]
Signed-off-by: Petr Štetiar <ynezz@true.cz>
3 years agopackages: kernel: add gpio-nxp-74hc153
Mauri Sandberg [Thu, 21 May 2020 06:59:52 +0000 (09:59 +0300)]
packages: kernel: add gpio-nxp-74hc153

NXP 74HC153 is a GPIO expander. Its original source cide sits in ar71xx
architecture tree. It has been slightly modified to get GPIO pin
configuration from the device tree rather than a MACH file.

 Changes to the source file:
  - Remove struct nxp_74hc153_config
  - in nxp_74hc153_probe(), fetch GPIO configuration from device tree
  - allow GPIO framework decide the base number by passing -1 to it
  - remove support for kernel versions below 4.5.0
  - add OF device compatibility string

 Create a package for inclusion in image.

References: https://lore.kernel.org/linux-gpio/545111184.50061.1615922388276@ichabod.co-bxl/
Signed-off-by: Mauri Sandberg <sandberg@mailfence.com>
[added link to driver usptreaming work in progress]
Signed-off-by: Petr Štetiar <ynezz@true.cz>
3 years agobcm4908: backport the latest bcm_sf2 commits
Rafał Miłecki [Mon, 22 Mar 2021 07:09:53 +0000 (08:09 +0100)]
bcm4908: backport the latest bcm_sf2 commits

1. CFP support for BCM4908
2. Upstream RGMII regs fix

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
3 years agobcm4908: backport DTS patch with Ethernet TX IRQ
Rafał Miłecki [Mon, 22 Mar 2021 06:40:38 +0000 (07:40 +0100)]
bcm4908: backport DTS patch with Ethernet TX IRQ

It allows bcm4908_enet Linux driver to work more efficiently.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
3 years agobcm4908: add MTD support for repartitioning
Rafał Miłecki [Mon, 22 Mar 2021 06:38:06 +0000 (07:38 +0100)]
bcm4908: add MTD support for repartitioning

It's required for sysupgrade which requires:
1. Flashing new firmware
2. Finding updated bootfs partition
3. Updating bootfs content

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
3 years agoath79: fix lan port display order for sitecom wlr-7100
Walter Sonius [Sun, 21 Mar 2021 19:06:33 +0000 (20:06 +0100)]
ath79: fix lan port display order for sitecom wlr-7100

Physical port order watched from the back of the device is:
4 / 3 / 2 / 1 / WAN which also matches corresponding leds.
This patch corrects LuCI switch webpage LAN port order.

Signed-off-by: Walter Sonius <walterav1984@gmail.com>
[improve commit title, fix sorting in 02_network]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agokernel: bump 5.4 to 5.4.106
John Audia [Wed, 17 Mar 2021 18:22:56 +0000 (14:22 -0400)]
kernel: bump 5.4 to 5.4.106

Ran update_kernel.sh in a fresh clone without any existing toolchains.

Manually rebased:
  bcm27xx/950-0993-xhci-quirks-add-link-TRB-quirk-for-VL805.patch
  layerscape/701-net-0231-enetc-Use-DT-protocol-information-to-set-up-the-port.patch

Build system: x86_64
Build-tested: ipq806x/R7800
Run-tested: ipq806x/R7800

No dmesg regressions, everything functional

Signed-off-by: John Audia <graysky@archlinux.us>
[remove accidental whitespace edit]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agoigmpproxy: remove package
Jan Pavlinec [Mon, 8 Mar 2021 16:33:46 +0000 (17:33 +0100)]
igmpproxy: remove package

Moved to packages repo because it was considered
non-essential for most router configurations.

Signed-off-by: Jan Pavlinec <jan.pavlinec@nic.cz>
[shorten commit title]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agoramips: add support for ELECOM WRC-1750GST2
INAGAKI Hiroshi [Wed, 17 Mar 2021 09:35:34 +0000 (18:35 +0900)]
ramips: add support for ELECOM WRC-1750GST2

ELECOM WRC-1750GST2 is a 2.4/5 GHz band 11ac (Wi-Fi 5) router, based on
MT7621A.

Specification:

- SoC : MediaTek MT7621A
- RAM : DDR3 256 MiB (NT5CC128M16JR-EK)
- Flash : SPI-NOR 32 MiB (MX25L25645GMI-08G)
- WLAN : 2.4/5 GHz 3T3R (2x MediaTek MT7615)
- Ethernet : 10/100/1000 Mbps x5
  - Switch : MediaTek MT7530 (SoC)
- LEDs/Keys : 4x/6x (2x buttons, 1x slide-switch)
- UART : through-hole on PCB
  - J4: 3.3V, GND, TX, RX, from ethernet port side
  - 57600n8
- Power : 12 VDC, 1.5 A

Flash instruction using factory image:

1. Boot WRC-1750GST2 normally with "Router" mode
2. Access to "http://192.168.2.1/" and open firmware update page
   ("ファームウェア更新")
3. Select the OpenWrt factory image and click apply ("適用") button
4. Wait ~120 seconds to complete flashing

MAC addresses:

LAN : 04:AB:18:xx:xx:23 (Factory, 0xE000 (hex))
WAN : 04:AB:18:xx:xx:24 (Factory, 0xE006 (hex))
2.4GHz : 04:AB:18:xx:xx:25 (Factory, 0x4    (hex))
5GHz : 04:AB:18:xx:xx:26 (Factory, 0x8004 (hex))

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agobase-files: add logging for configuration import
Florian Eckert [Mon, 1 Feb 2021 09:58:09 +0000 (10:58 +0100)]
base-files: add logging for configuration import

Make sysupgrade backup import more verbose.

Signed-off-by: Florian Eckert <fe@dev.tdt.de>
3 years agoramips: create shared DTSI for MT7620 Phicomm K2x series devices
Shiji Yang [Sun, 21 Mar 2021 08:45:24 +0000 (16:45 +0800)]
ramips: create shared DTSI for MT7620 Phicomm K2x series devices

Improve compatibility of the device tree include file. Now a new .dtsi
file will support both PSG1218A, PSG1218B and K2G.

Signed-off-by: Shiji Yang <yangshiji66@qq.com>
[improve commit title, rebase]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agoramips: increase SPI frequency for Phicomm series devices
Shiji Yang [Sun, 21 Mar 2021 09:51:00 +0000 (17:51 +0800)]
ramips: increase SPI frequency for Phicomm series devices

From many teardown image in the internet, I find Phicomm K1/k2 series use
Winbond W25Q64/W25Q128 or GigaDevice GD25Q64/GD25Q128 Flash chips. both of
them support 100+ MHz clock spi operate and fast-read instruction. PSG1218
with W25x or GD25x has been tested and it can run well in OpenWrt v19.07.

Signed-off-by: Shiji Yang <yangshiji66@qq.com>
[improve commit title]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agoramips: increase SPI frequency for HIWIFI HC5x61 devices
Shiji Yang [Sun, 21 Mar 2021 14:31:44 +0000 (22:31 +0800)]
ramips: increase SPI frequency for HIWIFI HC5x61 devices

HIWIFI HC5x61 devices support high speed spi clock up to 100+ MHz.
So set spi frequency to 80 MHz here (Due to frequency division the
real clock is 48 MHz).
I have tested HC5661 and it can run well in OpenWrt v19.07.

Signed-off-by: Shiji Yang <yangshiji66@qq.com>
[adjust commit title and wrap message]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agouhttpd: update to git HEAD
Hauke Mehrtens [Sun, 21 Mar 2021 21:34:13 +0000 (22:34 +0100)]
uhttpd: update to git HEAD

15346de client: Always close connection with request body in case of error

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
3 years agouhttpd: Execute uci commit and reload_config once
Hauke Mehrtens [Sat, 20 Mar 2021 12:26:20 +0000 (13:26 +0100)]
uhttpd: Execute uci commit and reload_config once

Instead of doing uci commit and reload_config for each setting do it
only once when one of these options was changed. This should make it a
little faster when both conditions are taken.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
3 years agouhttpd: Reload config after uhttpd-mod-ubus was added
Hauke Mehrtens [Sat, 20 Mar 2021 12:19:22 +0000 (13:19 +0100)]
uhttpd: Reload config after uhttpd-mod-ubus was added

Without this change the config is only committed, but the uhttpd daemon
is not reloaded. This reload is needed to apply the config. Without the
reload of uhttpd, the ubus server is not available over http and returns
a Error 404.

This caused problems when installing luci on the snapshots and
accessing it without reloading uhttpd.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
3 years agomediatek: linksys-e8450: remove left-overs from dtsi
Daniel Golle [Sun, 21 Mar 2021 13:32:46 +0000 (13:32 +0000)]
mediatek: linksys-e8450: remove left-overs from dtsi

There is an ASMedia ASM1480 PCIe switch found on mt7622-rfb1 and the
BPi-R64, allowing the user to switch between SATA and PCIe1 which share
the same pins on the SoC.
This chip is not present on the Linksys E8450, it doesn't have SATA.
Remove definitions for GPIO90 from DTSI to prevent it from being
copy&pasted or otherwise causing confusion.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
3 years agoprocd: update to git HEAD
Daniel Golle [Sun, 21 Mar 2021 13:31:51 +0000 (13:31 +0000)]
procd: update to git HEAD

 7ee4563 procd: Adding support to detect Pantavisor Container Platform

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
3 years agobcm27xx: bcm2711: correctly disable HW_RANDOM_BCM2835
Álvaro Fernández Rojas [Sun, 21 Mar 2021 11:39:47 +0000 (12:39 +0100)]
bcm27xx: bcm2711: correctly disable HW_RANDOM_BCM2835

It was removed in bac74aff5e36, but it should have been disabled.
More info: https://forum.openwrt.org/t/make-image-stops-at-random-number-generator-bcm27xx/91429/7

Fixes: bac74aff5e36 ("bcm27xx: bcm2711: disable HW_RANDOM_BCM2835")
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
3 years agobuild,json: fix duplicates in default_packages
Paul Spooren [Sun, 21 Mar 2021 09:23:34 +0000 (23:23 -1000)]
build,json: fix duplicates in default_packages

Calling without the DUMP=1 argument causes the target specific Makefile
to be "included" again which adds the target specific packages twice,
once on the actual run and once included from `include/target.mk`.

This led to duplicate package entries, causing confusion in downstream
projects using the generated JSON files.

While at it, apply `black` style to Python script.

Signed-off-by: Paul Spooren <mail@aparcar.org>
3 years agokernel: robustify dependencies in kmod-sound-hda-core
Tony Ambardar [Sun, 21 Mar 2021 06:30:44 +0000 (23:30 -0700)]
kernel: robustify dependencies in kmod-sound-hda-core

Dependency tracking for kmod-sound-hda-core is fragile. Enabling some sound
codecs (Realtek, Conexant, Sigmatel) implicitly adds a kmod-ledtrig-audio
dependency, while an enabled kmod-ledtrig-audio can be picked up through
enabling others (e.g. kmod-sound-hda-intel), and the behaviour can change
across kernel versions.

As kmod-ledtrig-audio is under 2KB, make it an unconditional dependency.

Fixes: a374b8f19051 ("kernel: 5.10: update sound modules")
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
3 years agoglibc: update to latest 2.33 commit
Hans Dedecker [Sat, 20 Mar 2021 20:00:05 +0000 (21:00 +0100)]
glibc: update to latest 2.33 commit

db32fc27e7 test-container: Always copy test-specific support files [BZ #27537]
79c6be6a0a nptl: Remove private futex optimization [BZ #27304]
f90d6b0484 pthread_once hangs when init routine throws an exception [BZ #18435]
dd8023c2ac elf: ld.so --help calls _dl_init_paths without a main map [BZ #27577]
ea5a537e87 elf: Always set l in _dl_init_paths (bug 23462)
64f6c287ad x86: Handle _SC_LEVEL1_ICACHE_LINESIZE [BZ #27444]
32b9280f1d io: Return EBAFD for negative file descriptor on fstat (BZ #27559)

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
3 years agolibnfnetlink: quote $(FPIC) on command line
Philip Prindeville [Fri, 19 Mar 2021 23:09:26 +0000 (17:09 -0600)]
libnfnetlink: quote $(FPIC) on command line

When $(FPIC) gets expanded on the command line (for instance
when setting environment variables for libtool, configure, or
make) we can't count on it not needing quoting (i.e. it could
contain multiple flags separated with spaces).

Signed-off-by: Philip Prindeville <philipp@redfish-solutions.com>
3 years agoumdns: add syscalls needed on Aarch64
Daniel Golle [Fri, 19 Mar 2021 22:32:12 +0000 (22:32 +0000)]
umdns: add syscalls needed on Aarch64

Now that ujail supports seccomp also on Aarch64, add missing syscall
'fstat' to the list of allowed syscalls.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
3 years agoprocd: update to git HEAD
Daniel Golle [Fri, 19 Mar 2021 22:30:42 +0000 (22:30 +0000)]
procd: update to git HEAD

Enable seccomp features on Aarch64.

 3e88c6f jail/seccomp: add support for aarch64
 c23d8bf trace: fix build on aarch64

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
3 years agofirewall4: introduce package
Jo-Philipp Wich [Thu, 22 Oct 2020 14:28:51 +0000 (16:28 +0200)]
firewall4: introduce package

This commit introduces firewall4, an nftables based reimplementation of the
UCI iptables firewall.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
3 years agoucode: introduce package
Jo-Philipp Wich [Thu, 22 Oct 2020 14:09:56 +0000 (16:09 +0200)]
ucode: introduce package

This commit introduces ucode, a tiny template processor utilizing a syntax
closely resembling ECMA Script for embedded script blocks.

Signed-off-by: Jo-Philipp Wich <jo@mein.io>
3 years agokernel: Add missing kernel config options
Hauke Mehrtens [Fri, 19 Mar 2021 17:48:32 +0000 (18:48 +0100)]
kernel: Add missing kernel config options

These two options are missing when building the x86 target.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
3 years agokernel: Reorder kernel configuration options
Hauke Mehrtens [Fri, 19 Mar 2021 17:35:23 +0000 (18:35 +0100)]
kernel: Reorder kernel configuration options

Reorder the options and remove double entries.
This was generated by running this:
$ ./scripts/kconfig.pl '+' target/linux/generic/config-5.10 /dev/null > target/linux/generic/config-5.10.new
$ mv target/linux/generic/config-5.10.new target/linux/generic/config-5.10
$ ./scripts/kconfig.pl '+' target/linux/generic/config-5.4 /dev/null > target/linux/generic/config-5.4.new
$ mv target/linux/generic/config-5.4.new target/linux/generic/config-5.4

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
3 years agoarchs38: Add CONFIG_HZ=100
Hauke Mehrtens [Fri, 19 Mar 2021 16:59:12 +0000 (17:59 +0100)]
archs38: Add CONFIG_HZ=100

This kernel config option was missing and resulted in a question when
building.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
3 years agokernel: add exfat again
Rosen Penev [Tue, 16 Mar 2021 22:46:03 +0000 (15:46 -0700)]
kernel: add exfat again

With kernel 5.10, exfat is out of staging and in tree.

Added small hack to make it work with kernel 5.4 as well.

Added removed config options for 5.4 to generic config.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
[Set CONFIG_EXFAT_ config options to default values]
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
3 years agopackage/exfat: remove
Rosen Penev [Tue, 16 Mar 2021 22:46:02 +0000 (15:46 -0700)]
package/exfat: remove

Now that kernel 5.10 is in tree, this can be removed.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
3 years agomvebu: enable WRT1900AC v1 and WRT32X for buildbots
Tad Davanzo [Tue, 9 Feb 2021 22:42:54 +0000 (17:42 -0500)]
mvebu: enable WRT1900AC v1 and WRT32X for buildbots

Kernel size limits have been dealt with.
Effective revert of a1eb2c46 and ac9730c4.

Signed-off-by: Tad Davanzo <tad@spotco.us>
3 years agomvebu: venom resize kernel to 6MB
Tad Davanzo [Sat, 6 Feb 2021 03:42:47 +0000 (22:42 -0500)]
mvebu: venom resize kernel to 6MB

venom has a 3MB kernel partition as specified by the DTS.
3MB is not sufficient for building with many kernel modules or newer
kernel versions.

venom uboot however as set from factory will load up to 6MB.
This can be observed by looking a uboot log:
NAND read: device 0 offset 0x900000, size 0x600000
6291456 bytes read: OK
and from uboot environment variables:
$ fw_printenv | grep "priKernSize";
priKernSize=0x0600000

Resize the root partitions from 120MB to 117MB to let kernel expand
into it another 3MB.
And set kernel target size to 6MB.

Lastly set the kernel-size-migration compatibility version on venom to
prevent sysupgrading without first reinstalling from a factory image.

Signed-off-by: Tad Davanzo <tad@spotco.us>
3 years agomvebu: mamba resize kernel to 4MB
Tad Davanzo [Sat, 6 Feb 2021 03:38:03 +0000 (22:38 -0500)]
mvebu: mamba resize kernel to 4MB

mamba has a 3MB kernel partition as specified by the DTS.
3MB is not sufficient for building with many kernel modules or newer
kernel versions.

mamba uboot however as set from factory will load up to 4MB.
This can be observed by looking a uboot log:
NAND read: device 0 offset 0xa00000, size 0x400000
4194304 bytes read: OK
and from uboot environment variables:
$ fw_printenv | grep "pri_kern_size";
pri_kern_size=0x400000

Resize the root partitions from 37MB to 36MB to let kernel expand
into it another 1MB.
And set kernel target size to 4MB.

Lastly add a compatibility version message: kernel-size-migration.
And set it on mamba to prevent sysupgrading without first reinstalling from
a factory image.

Signed-off-by: Tad Davanzo <tad@spotco.us>
3 years agoiproute2: separate tc into tiny and full variants
Tony Ambardar [Tue, 2 Mar 2021 01:14:50 +0000 (17:14 -0800)]
iproute2: separate tc into tiny and full variants

This change was investigated previously [1] but not deemed necessary. With
the recent addition [2] of modern BPF loader support, however, tc gained
dependencies on libelf and libbpf, with a larger installation footprint.

Similar to ip-tiny/ip-full, split tc into tc-full and tc-tiny variants,
where the latter excludes the eBPF loader, uses a smaller executable, and
avoids libelf and libbpf package dependencies. Both variants provide the
'tc' virtual package, with tc-tiny as the default.

The previous tc package included a loadable module for iptables actions.
Separate this out into a common package, tc-mod-iptables, which both
variants depend on. Some package sizes on mips_24kc:

Before:
  148343  tc_5.11.0-1_mips_24kc.ipk

After:
  144833  tc-full_5.11.0-2_mips_24kc.ipk
  138430  tc-tiny_5.11.0-2_mips_24kc.ipk  (and no libelf or libbpf)
    4115  tc-mod-iptables_5.11.0-2_mips_24kc.ipk

Also fix up some Makefile indentation.

[1] https://github.com/openwrt/openwrt/pull/1627#issuecomment-447619962
[2] b048a305a3d3 ("iproute2: update to 5.11.0")

Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
3 years agokernel/modules: relocate teql hotplug from iproute2 to kmod-sched
Tony Ambardar [Tue, 2 Mar 2021 04:59:04 +0000 (20:59 -0800)]
kernel/modules: relocate teql hotplug from iproute2 to kmod-sched

The link equalizer sch_teql.ko of package kmod-sched relies on a hotplug
script historically included in iproute2's tc package. In previous
discussion [1], consensus was the hotplug script is best located together
with the module in kmod-sched, but this change was deferred at the time.

Relocate the hotplug script now. This change also simplifies adding a tc
variant for minimal size with reduced functionality.

[1] https://github.com/openwrt/openwrt/pull/1627#issuecomment-447923636

Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
3 years agoiproute2: add missing limits.h includes
Tony Ambardar [Sat, 13 Mar 2021 04:58:28 +0000 (20:58 -0800)]
iproute2: add missing limits.h includes

This patch has been submitted upstream to fix an error reported by a few
users. One instance seen using gcc 10.2.0, binutils 2.35.1 and musl 1.1.24:

bpf_glue.c: In function 'get_libbpf_version':
bpf_glue.c:46:11: error: 'PATH_MAX' undeclared (first use in this function);
did you mean 'AF_MAX'?
   46 |  char buf[PATH_MAX], *s;
      |           ^~~~~~~~
      |           AF_MAX

Reported-by: Rui Salvaterra <rsalvaterra@gmail.com>
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
3 years agokernel: add missing config symbols, fix typos
Tony Ambardar [Thu, 18 Mar 2021 07:06:47 +0000 (00:06 -0700)]
kernel: add missing config symbols, fix typos

Update 5.10 generic config to add missing options prompting during builds,
when setting CONFIG_ALL_KMODS. Refresh the x86 subtargets to drop symbols
added to the generic config.

Also fix a typo in 5.4/5.10 generic configs: "CONFIG_CONFIG_KVM_AMD_SEV".

Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
3 years agokernel: 5.10: update sound modules
Tony Ambardar [Wed, 17 Mar 2021 10:12:33 +0000 (03:12 -0700)]
kernel: 5.10: update sound modules

Update file paths for kmod-sound-hda-intel and reflect new dependency of
kmod-sound-hda-core on kmod-ledtrig-audio.

Reported-by: Javier Marcet <javier@marcet.info>
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
3 years agokernel: add kmod-ledtrig-audio
Tony Ambardar [Wed, 17 Mar 2021 11:56:23 +0000 (04:56 -0700)]
kernel: add kmod-ledtrig-audio

This allows LEDs to be controlled by audio drivers.

Signed-off-by: Tony Ambardar <itugrok@yahoo.com>