After years of trying to find the reason for random kernel crashes
while both CPU and SATA are under load it has been found.
Some odd commented-out #defines in kref's single-port driver [1] which
were copied from the vendor driver made me develop a theory:
The IO-mapped memory area for DMA descriptors apparetly got some holes
just before the alignment boundaries.
This feels like an off-by-one bug in the hardware or maybe those fields
are used internally by the SATA controller's firmware.
Whatever the cause is: they cannot be used and trying to use them
results in reading back unexpected stuff and ends up with oopsing
Unable to handle kernel paging request at virtual address d085c004
Work around the issue by reducing the area used for bmdma descriptors.
This reduces SATA performance (iops) quite a bit, but finally makes
things work reliably. Possibly one could optimize this much more by
really just skipping the holes in that memory area -- however, that
seems to be non-trivial with the driver and libata in it's current form
(suggestions are welcome).
The 'proper' way to have good SATA performance would be to make use of
the hardware RAID features (one can use the JBOD mode to access even
just a single disc transparently through the RAID controller integrated
in the SATA host instead of accessing the SATA ports 'raw' as we do
now).
[1]: https://github.com/kref/linux-oxnas/blob/master/drivers/ata/sata_oxnas.c#L25
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Most newer targets have been converted to consistently use
vendor_model scheme for device definitions/image names, ox820 is
using it as well, so let's just convert ox810se for consistency.
While at it, use generic setup for DEVICE_DTS and add SUPPORTED_DEVICES.
The latter have been introduced for ox820 already in
cf7896117b ("oxnas: enable image metadata by setting SUPPORTED_DEVICES")
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Many target use a repetitive if-include scheme for their subtarget
image files, though their names are consistent with the subtarget
names.
This patch removes these redundant conditions and just uses the
variable for the include where the target setup allows it.
For sunxi, this includes a trivial rename of the subtarget image
Makefiles.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
CONFIG_VLAN_8021Q was explicitely disabled in oxnas kernel config.
Don't do that, so VLANs can be used on the target.
Fixes: dcc34574ef ("oxnas: bring in new oxnas target")
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
This service file has been misplaced from the very beginning.
Fixes: dcc34574ef ("oxnas: bring in new oxnas target")
Signed-off-by: Sungbo Eo <mans0n@gorani.run>
* install kmod-hwmon-drivetemp by default
* wire up thermal zone
* fix fan GPIO polarity
* fix i2c-gpio GPIO_OPEN_DRAIN
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
drivers/ata/sata_oxnas.c: In function 'sata_oxnas_port_irq':
drivers/ata/sata_oxnas.c:2126:25: warning: left shift count >= width of type [-Wshift-count-overflow]
if (ap->qc_active & (1 << ATA_TAG_INTERNAL)) {
^~
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
This time DTS fix, again from Sungbo Eo <mans0n@gorani.run>
ARM: dts: oxnas: Fix clear-mask property
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Sungbo Eo <mans0n@gorani.run> submitted another patch fixing an error
on reboot:
irqchip/versatile-fpga: Apply clear-mask earlier
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Sungbo Eo <mans0n@gorani.run> posted a patch fixing the long-standing
reboot problem on the OXNAS OX820 platform:
irqchip/versatile-fpga: Handle chained IRQs properly
It got queued for 5.7. Import it to oxnas target patches for now.
Fixes: b4917fa907 ("oxnas: fix oxnas-rps-timer dt-match")
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Pogoplug V3/Pro has an interanl SATA port. To use it, DTS sata node should be
enabled, and kmod-ata-oxnas-sata package needs to be installed.
Fixes: FS#2542
Signed-off-by: Sungbo Eo <mans0n@gorani.run>
This new symbol popped up in few places. Disable it in generic config.
Signed-off-by: Tomasz Maciej Nowak <tomek_n@o2.pl>
[fixed merge conflict in generic/config-5.4]
Signed-off-by: Petr Štetiar <ynezz@true.cz>
There is no such role as target maintainer anymore, one should always
send corresponding changes for the review and anyone from the commiters
is allowed to merge them or eventually use the hand break and NACK them.
Lets make it clear, that it is solely a community doing the maintenance
tasks.
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Acked-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Acked-by: Piotr Dymacz <pepe2k@gmail.com>
As it seeems these patches are for OpenWrt and they are not going to be merged
to mainline, let's merge them to improve maintainability.
Signed-off-by: Sungbo Eo <mans0n@gorani.run>
Follow the guideline suggested in e417ff88f1 ("ramips: harmonize line breaks
in image Makefiles").
While at it, fix typos in the device title of Akitio MyCloud.
Signed-off-by: Sungbo Eo <mans0n@gorani.run>
Currently kmod-hwmon-* will not get into images unless kmod-hwmon-core is added
to DEVICE_PACKAGES as well. By changing the dependencies from "depends on" to
"select", we do not have the issue anymore.
Furthermore, we can remove most occurrences of the package from DEVICE_PACKAGES
and similar variables, as it is now pulled by dependent modules such as:
- kmod-hwmon-gpiofan
- kmod-hwmon-lm63
- kmod-hwmon-lm75
- kmod-hwmon-lm85
- kmod-hwmon-lm90
Signed-off-by: Sungbo Eo <mans0n@gorani.run>
[do not touch ar71xx, adjust line wrapping]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Run-tested 5.4 on Shuttle KD20 for some days now, everything seems
fine so far. Let's have snapshot builds based on 5.4.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
DTS variable is not used in any device. Just drop it.
UBIFS_OPTS is not device-dependent here. Remove it from DEVICE_VARS.
Signed-off-by: Sungbo Eo <mans0n@gorani.run>
patches-4.14/340-oxnas-pcie.patch was adapted for kernel 5.4 by
accident. Revert that change to restore 4.14 builds.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Add 5.4 kernel version as a new testing kernel option.
Run-tested on Shuttle KD20, seems to work just as well as kernel 4.14.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Enabling legacy PTYs causes problems with procd-hotplug.
And as this is a headless target, no need to have virtual terminals.
Remove corresponding kernel config options, they are disabled in
generic kernel config.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
PogoPlug Pro comes with AzureWave AW-NE762H PCIe module containing
Ralink's Rt3090 chip supported by the rt2x00 driver.
Install the driver as well as wpad-basic.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Move PCIe controller outside down to SoC level to avoid resource
mapping problems.
Also add more detailed error handling when mapping registers.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
They should be relative to apb-bridge@47000000 rather than to the
pcie-controller@c00000 inside it.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Refactor pcie-oxnas to have shared resources in syscon and new pcie-phy
driver. Hopefully this revives PCIe...
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
