Following errors were seen in the past on imx6 when using serial:
[ 22.617622] imx-uart 2020000.serial: DMA transaction error.
[ 22.623228] imx-uart 2020000.serial: DMA transaction error.
[ 22.628826] imx-uart 2020000.serial: DMA transaction error.
[ 22.648951] imx-uart 2020000.serial: DMA transaction error.
[ 22.654558] imx-uart 2020000.serial: DMA transaction error.
[ 22.660156] imx-uart 2020000.serial: DMA transaction error.
Which is the reason why DMA for the serial ports
got disabled in commits:
efb362cd93 ("imx6: disable dma on uart")
3b4241071d ("imx6: disable UART dma")
As indicated on mailinglist discussion, the cause seems to be
the usage of very old SDMA firmware which is present in the soc:
[ 0.624302] imx-sdma 20ec000.sdma: Direct firmware load for imx/sdma/sdma-imx6q.bin failed with error -2
[ 0.624318] imx-sdma 20ec000.sdma: Falling back to user helper
[ 64.531607] imx-sdma 20ec000.sdma: external firmware not found, using ROM firmware
This patch adds the new firmware binary. (2196 bytes)
It is required to embed the binary into the kernel image, as it
gets loaded very early in the boot process where the rootfs is not
available yet:
[ 0.622966] imx-sdma 20ec000.sdma: loaded firmware 3.3
Extended testing shows that the DMA errors are not seen anymore
when using this newer firmware version.
Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
starting from upstream commit 577b4eb23811 ("ubi: Reject MLC NAND")
it is not allowed to use UBI and UBIFS on a MLC flavoured NAND flash chip. [1]
According to David Oberhollenzer [2]:
The real problem is that on MLC NAND, pages come in pairs.
Multiple voltage levels inside a single, physical memory cell are used to
encode more than one bit. Instead of just having pages that are twice as big,
the flash exposes them as two different pages. Those pages are usually not
ordered sequentially either, but according to a vendor/device specific
pairing scheme.
Within OpenWrt, devices utilizing this type of flash,
combined with UBI(fs) will be bricked when a user upgrades
from 17.01.4 to a newer version as the MLC will be refused.
As these devices are currently advertised as supported by OpenWrt,
we should at least maintain the original state during the lifecycle
of the current releases.
Support can be gracefully ended when a new release-branch is created.
Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
Acked-by: Hauke Mehrtens <hauke@hauke-m.e>
[1] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v4.14.77&id=577b4eb23811dfc8e38924dc476dbc866be74253
[2] https://lore.kernel.org/patchwork/patch/920344/
This just moves patch to use 0xx prefix and includes maintainer's s-o-b.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
(cherry picked from commit 9b385b2496)
Instead of reverting whole commit it's enough to just revert a single
line change. It seems the real problem with the regressing commit was a
bump of read chunk size. Switching back to 256 B chunks is enough to fix
the problem/regression.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
(cherry picked from commit 92de28b751)
Refreshed all patches.
Altered patches:
- 666-Add-support-for-MAP-E-FMRs-mesh-mode.patch
New symbol for arm targets:
- HARDEN_BRANCH_PREDICTOR
Compile-tested on: cns3xxx, imx6, x86_64
Runtime-tested on: cns3xxx, imx6, x86_64
Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
Tested-by: Stijn Tintel <stijn@linux-ipv6.be>
In commit 9e1530b2a3 ("kernel: bump 4.9 to 4.9.117 for 18.06") [1], the following patch for removed:
- 403-mtd_fix_cfi_cmdset_0002_status_check.patch
This patch contained fixes for both write and erase functions.
While the chip-detects for erase got fixed upstream [2],
some modifications are still required, even with the fixes applied.
Not doing so results in following errors seen:
Collected errors:
* pkg_write_filelist: Failed to open //usr/lib/opkg/info/luci-lib-ip.list: I/O error.
* opkg_install_pkg: Failed to extract data files for luci-lib-ip. Package debris may remain!
* opkg_install_cmd: Cannot install package luci-ssl.
* opkg_conf_write_status_files: Can't open status file //usr/lib/opkg/status: I/O error.
[ 0.780920] jffs2: version 2.2 (NAND) (SUMMARY) (LZMA) (RTIME) (CMODE_PRIORITY) (c) 2001-2006 Red Hat, Inc.
[ 8.406396] jffs2: notice: (415) jffs2_build_xattr_subsystem: complete building xattr subsystem, 0 of xdatum (0 unchecked, 0 orphan) and 0 of xref (0 dead, 0 orphan) found.
[ 8.423476] mount_root: switching to jffs2 overlay
[ 270.902671] jffs2: Write of 1989 bytes at 0x005ce6f8 failed. returned -5, retlen 962
[ 270.931965] jffs2: Write of 1989 bytes at 0x005ceec0 failed. returned -5, retlen 0
[ 270.939631] jffs2: Not marking the space at 0x005ceec0 as dirty because the flash driver returned retlen zero
[ 270.950397] jffs2: Write of 68 bytes at 0x005ceec0 failed. returned -5, retlen 0
[ 270.957838] jffs2: Not marking the space at 0x005ceec0 as dirty because the flash driver returned retlen zero
[ 270.968584] jffs2: Write of 68 bytes at 0x005ceec0 failed. returned -5, retlen 0
[ 270.976027] jffs2: Not marking the space at 0x005ceec0 as dirty because the flash driver returned retlen zero
[ 270.986735] jffs2: Write of 68 bytes at 0x005ceec0 failed. returned -5, retlen 0
[ 270.994225] jffs2: Not marking the space at 0x005ceec0 as dirty because the flash driver returned retlen zero
[1] https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=fec8fe806963c96a6506c2aebc3572d3a11f285f
[2] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v4.9.133&id=a0239d83e1cb60de5e78452d4708c083b9e3dcbe
Fixes: 9e1530b2a3 ("kernel: bump 4.9 to 4.9.117 for 18.06")
Signed-off-by: Fabio Bettoni <fbettoni@gmail.com>
Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
Compaction is the only memory management component to form high order (larger
physically contiguous) memory blocks reliably. The page allocator relies on
compaction heavily and the lack of the feature can lead to unexpected OOM
killer invocations for high order memory requests. You shouldn't disable this
option unless there really is a strong reason for it.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Michal Hrusecky <michal.hrusecky@nic.cz>
The install_bin from /lib/upgrade/common.sh is no longer creating the
symlinks when a secondary parameter is added. But the fw_setenv program was
always copied this way to the ramdisk for the upgrade.
Instead, just install fw_setenv and let install_bin handle the detection of
the required dependencies.
Fixes: 438dcbfe74 ("base-files: automatically handle paths and symlinks for RAMFS_COPY_BIN")
Signed-off-by: Sven Eckelmann <sven.eckelmann@openmesh.com>
Buttons of AVM FritzBox 4020 are incorrectly flagged as active high.
This was an oversight as RFKill button was working as expected even
with incorrectly flagged GPIO.
Signed-off-by: David Bauer <mail@david-bauer.net>
(cherry picked from commit cd02d4faf9)
The sysupgrade image failed the check due to the wrong string in the
supported devices. This patch provides the correct name by dropping the
SUPPORTED_DEVICES to use the default generated name.
Signed-off-by: Steffen Förster <steffen@chemnitz.freifunk.net>
[drop the SUPPORTED_DEVICES, the old name was never used in a release]
Signed-off-by: Mathias Kresin <dev@kresin.me>
When building using the multiple devices option with per-device root
filesystem, only the meta package mt76 is omitted but not the
dependencies selected by the package.
Explicitly exclude all 3 mt76 packages, plus the metapackage.
Otherwise, these modules will be included in the build, wasting
a few hundred kilobytes.
Signed-off-by: Joseph C. Lehner <joseph.c.lehner@gmail.com>
[mention the root cause of the issue in the commit message]
Signed-off-by: Mathias Kresin <dev@kresin.me>
This fixes regression introduced in kernel 4.14 and makes bcm53xx revert
obsolete.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
(cherry picked from commit 43d36606d6)
PCI paths of the WLAN devices have changed between kernel 4.4 and 4.9;
migrate config so existing wifi-iface definitions don't break.
This is implemented as a hotplug handler rather than a uci-defaults script
as the migration script must run before the 10-wifi-detect hotplug handler.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
(cherry picked from commit b452af23a8)
When checking the outcome of the PHY autonegotiation status, at803x
currently returns false in case the SGMII side is not established.
Due to a hardware-bug, ag71xx needs to fixup the SoCs SGMII side, which
it can't as it is not aware of the link-establishment.
This commit allows to ignore the SGMII side autonegotiation status to
allow ag71xx to do the fixup work.
Signed-off-by: David Bauer <mail@david-bauer.net>
(cherry picked from commit 4e39e213af)
The QCA955X is affected by a hardware bug which causes link-loss of the
SGMII link between SoC and PHY. This happens on change of link-state or
speed.
It is not really known what causes this bug. It definitely occurs when
using a AR8033 Gigabit Ethernet PHY.
Qualcomm solves this Bug in a similar fashion. We need to apply the fix
on a per-device base via platform-data as performing the fixup work will
break connectivity in case the SGMII interface is connected to a Switch.
This bug was first proposed to be fixed by Sven Eckelmann in 2016.
https://patchwork.ozlabs.org/patch/604782/
Based-on-patch-by: Sven Eckelmann <sven.eckelmann@open-mesh.com>
Signed-off-by: David Bauer <mail@david-bauer.net>
(cherry picked from commit f4f99ec973)
The changed dictionary size leads to a different LZMA header which breaks
sysupgrade image magic checkibng on at least some RT288x boards.
Since the commit message only mentions testing on MT7621 and since the
change appears to break at least one other ramips subtarget, do not take
any chances and restrict the size limitation to only MT7621.
Fixes FS#1797
Fixes 09b6755946 ("ramips: limit dictionary size for lzma compression")
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit 77e2bccde8)
Netgear WNR612v2 flashed with recent OpenWrt builds suffers from kernel
panic at boot during wireless chip initialization, making device
unusable:
ath: phy0: Ignoring endianness difference in EEPROM magic bytes.
ath: phy0: Enable LNA combining
CPU 0 Unable to handle kernel paging request at virtual address 1000fee1, epc == 801d08f0, ra == 801d0d90
Oops[#1]:
CPU: 0 PID: 469 Comm: kmodloader Not tainted 4.9.120 #0
[ ... register dump etc ... ]
Kernel panic - not syncing: Fatal exception
Rebooting in 1 seconds..
This simple patch fixes above error. It keeps LED table in memory after
kernel init phase for ath9k driver to operate correctly (__initdata
removed).
Also, another bug is fixed - correct array size is provided to function
that adds platform LEDs (this device has only 1 connected to Wifi chip)
preventing code from going outside array bounds.
Fixes: 1f5ea4eae4 ("ar71xx: add correct named default wireless led by using platform leds")
Signed-off-by: Michal Cieslakiewicz <michal.cieslakiewicz@wp.pl>
[trimmed commit message]
Signed-off-by: Mathias Kresin <dev@kresin.me>
The NBG6616 shares a config symbol with the NBG6716. It was accidentally
removed from the config when the ar71xx-tiny target was split off.
Fixes: 0cd5e85e7a ("ar71xx: create new ar71xx/tiny subtarget for 4MB flash devices")
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
(cherry picked from commit a4f4ddba61)
The device was not included in the /etc/board.d/02_network file, so
the network wouldn't be properly set up on boot.
Signed-off-by: Roger Pueyo Centelles <roger.pueyo@guifi.net>
The former used compatibles aren't defined anywhere and aren't used by
the devicetree source files including them.
Signed-off-by: Mathias Kresin <dev@kresin.me>
According to abbfcc8525 ("ramips: add support for GL-inet
GL-MT300N-V2") the board has a MediaTek MT7628AN. Change the SoC
compatible to match the used hardware.
Signed-off-by: Mathias Kresin <dev@kresin.me>
RT5350 neither have rgmii nor a mdio pinmux group. MT7628an doesn't
have a jtag group. Having these groups defined might cause a boot
panic.
The pin controller fails to initialise for kernels > 4.9 if invalid
groups are used. If a subsystem references a pin controller
configuration node, it can not find this node and errors out. In worst
case it's the SPI driver which errors out and we have no root
filesystem to mount.
Signed-off-by: Mathias Kresin <dev@kresin.me>
The intention of 967b6be118 ("ar8327: Add workarounds for AR8337
switch") was to remove the register fixups for AR8337. But instead they
were removed for AR8327.
The RGMII RX delay is forced even if the port is used as phy instead of
mac, which results in no package flow at least for one board.
Fixes: FS#1664
Signed-off-by: Mathias Kresin <dev@kresin.me>
Since kernel 4.10 commit 61e84623ace3 ("net: centralize net_device
min/max MTU checking"), the range of mtu is [min_mtu, max_mtu], which
is [68, 1500] by default.
It's necessary to set a max_mtu if a mtu > 1500 is supported.
Tested-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
Signed-off-by: Mathias Kresin <dev@kresin.me>
