Optimize the compiler output for larger cache blast cases that are
common for DMA-based networking.
On ar71xx, I measured a routing throughput increase of ~8%
Signed-off-by: Ben Menchaca <ben.menchaca@qca.qualcomm.com>
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
This bumps the 4.4. kernel in master to 4.4.119.
Includes more Meltdown & Spectre mitigation.
* Refresh patches.
* Refresh x86/config for RETPOLINE.
* Deleted 8049-PCI-layerscape-Add-fsl-ls2085a-pcie-compatible-ID.patch (accepted upstream)
* Deleted 8050-PCI-layerscape-Fix-MSG-TLP-drop-setting.patch (accepted upstream)
* Deleted 650-pppoe_header_pad.patch (does not apply anymore (code was replaced)).
Bumps from 4.4.112 to 4.4.115 were handled by Kevin Darbyshire-Bryant.
Compile-tested on: ar71xx & oxnas.
Signed-off-by: Stijn Segers <foss@volatilesystems.org>
Tested-by: Rosen Penev <rosenp@gmail.com>
This patch bumps the 4.14 kernel to .23.
- Refreshed patches.
- Deleted bcm53xx/patches-4.14/089-PCI-iproc-Fix-NULL-pointer-dereference-for-BCMA.patch. Has been accepted upstream.
- Deleted generic/pending-4.14/821-usb-Remove-annoying-warning-about-bogus-URB.patch. The upstream URB code was changed,
the patch no longer applies. I discussed this with the patch author and removed it for now, we'll see how it goes.
Compile-tested on: ramips/mt7621, x86/64
Run-tested on: ramips/mt7621, x86/64
Signed-off-by: Stijn Segers <foss@volatilesystems.org>
When CGROUPS is enabled the new option CONFIG_CGROUP_NET_CLASSID is
selectable and not handled.
Add this option to the 4.14 kernel configuration.
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
Somewhere in the kernel 4.9 migration process it looks like this patch
was broken. The update should be done when a device is added to the
bridge and when it gets removed, currently it is only done on the remove
path twice which does not make any sense.
I do not have a setup to test this now so this is only compile tested.
Fixes: f791fb4af4 ("kernel: add linux 4.9 support")
Fixes: b3f95490b9 ("kernel: generic: Add kernel 4.14 support")
Reported-by: Andrey Jr. Melnikov <temnota.am@gmail.com>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
This makes it possible to add an iptables rule that offloads routing/NAT
packet processing to a software fast path. This fast path is much
quicker than running packets through the regular tables/chains.
Requires Linux 4.14
Signed-off-by: Felix Fietkau <nbd@nbd.name>
This only works with nftables for now, iptables support will be added
later. Includes a number of related upstream nftables improvements to
simplify backporting follow-up changes
Signed-off-by: John Crispin <john@phrozen.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Add support to allow for per switch port VLAN priority (PCP) bits
for the ar8327/8337 chip using the swconfig utility.
Tested on Netgear R7800
Signed-off-by: Tan Hong Hui <hhtan72@yahoo.com>
This issue is also present in kernel 4.9 starting from 4.9.71
Adapted the patch, as the fixed function is in another location here.
Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
- removed upstreamed patches
- 0901-spansion_nand_id_fix.patch is disabled, not clear if it's needed
Signed-off-by: Roman Yeryomin <roman@advem.lv>
Signed-off-by: John Crispin <john@phrozen.org>
sysfs attributes 'port_mask' & 'speed_mask' held locks whilst doing
mundane tasks such as sprintf. Refactor code to reduce length of time
locks are held unnecessarily.
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
Add sysfs 'mode' attribute to swconfig controlled LEDs.
swconfig 'link state' LEDs blink in the presence of port traffic. This
behaviour becomes more obvious as switches start to support
get_port_stats() e.g. commits 0369e35891,
3056d09b40,
4ddbc43cc1,
4d8a66d934.
This blinking can be confusing/distracting if the switch has other LEDs
used to indicate traffic. Provide a 'mode' sysfs attribute that
controls the blink on traffic behaviour.
mode - either "none" (LED is off) or a space separated list of one or more:
link: LED's normal state reflects whether the link is up (has carrier) or not
tx: LED blinks on transmitted data
rx: LED blinks on receive data
Note that 'link' considers any port speed mask that may be applicable.
e.g. if an LED is configured to indicate 1Gbit link speed and mode is
set to 'link rx tx' but the port is connected at 100Mbit then the LED
will not light or blink. A mode of 'tx rx' will blink in the presence of
traffic only if the port matches the rate (if configured)
This maintains compatibility with existing behaviour.
Attribute is 'link tx rx' by default for backwards compatible behaviour.
Many thanks to Thibaut Varene for providing a more sensible led_event
routine after I had mangled the original, and other coding style hints.
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
Acked-by: Thibaut VARENE <hacks@slashdirt.org>
Instead of creating an ubifs file system with format version 5 by
default on empty UBI volumes use the older format 4 by default. This
will make it possible to mount these file systems also on older kernel
versions.
When a user wants to do a sysupgrade from kernel 4.14 to kernel 4.9 the
old kernel has to read the file system created by the more recent kernel
which currently does not work for ubifs.
This fixes the problem by creating file systems which are compatible
with older kernel versions by default.
Kernel 4.14 will still be able to read and write UBI FS file system
version 5, it will just not be used when a ubifs partition is created
implicitly on an empty UBI volume.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
The D-Link devices with JBOOT bootloader use their own kernel
image header (stag + sch2 headers).
This driver find jImage header and set rootfs start after kernel file.
Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
This config option was renamed in upstream Linux commit 681bec0367
("tracing: Rename update the enum_map file")
Reported-by: Rosen Penev <rosenp@gmail.com>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Locally generated packets weren't forwarded to the isolated interfaces in a
bridge. Isolation should only prevent the flooding of incomming packets to
other interfaces in the bridge.
Signed-off-by: Daniel Danzberger <daniel@dd-wrt.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
When CGROUPS is enabled the new option CONFIG_CGROUP_NET_CLASSID is
selectable and not handled.
Add this option to the 4.9 kernel configuration.
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
This replaces the current patches used to make the kernel headers
compatible with musl with the version which was accepted upstream. This
is included in upstream kernel 4.15.
This was compile tested with iproute2 build on all supported kernel
versions with musl and one one with glibc.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
nftables 0.8.1 generates some new commands which will not work without
this on big endian systems. This patch is included in Linux 4.11 and
later.
My rule matching a TCP port was not working:
nft add rule ip foo bar ct state new tcp dport 22 accept
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
The KEXEC_FILE symbol exists for X86 since kernel 3.17, and since 4.10
for PPC64. Add it to x86/config-4.9 and to generic/config-4.14.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
Several new DRM symbols that were introduced after 4.9 are missing in
the generic config for 4.14, so add them.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
While working on a new target (meson), the kernel build failed due to
missing DRM_DEBUG_MM_SELFTEST symbol. This can potentially happen on all
targets that enable DRM drivers in the kernel config or via kmod
packages, so add it to the generic config and remove it from x86
subtarget configs, together with DRM_DEBUG_MM.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
Add a bunch of missing configuration symbols found while building
armvirt for 4.14 after re-synchronization of the configuration between
4.9 and 4.14.
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
This backports upstream support for "compatible" DT property set for the
"partitions" subnode of flash node. It allows specifying how partitions
should be created/parsed. Right now only "fixed-partitions" is
supported.
It should eventually replace our downstream "linux,part-probe" solution.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Multicast routing support is not needed in most setups, and increases the
size of the kernel considerably (>10K after LZMA). Add a config switch to
allow disabling it.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
The underlying issue breaking Spansion flash has been fixed with "mtd: spi-nor:
wait until lock/unlock operations are ready" and "mtd: spi-nor: wait for SR_WIP
to clear on initial unlock", so we can support unlocking for Winbond flash
again.
This is necessary to have writable flash on certain UBNT devices with some
bootloader versions.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Rename unwinder config symbols to match upstream changes.
Refresh patches.
Update patch that no longer applies: 202-reduce_module_size.patch
Also enable CONFIG_PAGE_TABLE_ISOLATION. This feature was backported
from 4.15 to the 4.14 stable series. It is enabled by default, so enable
it in OpenWrt as well.
Compile-tested on x86/64.
Runtime-tested on x86/64.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
While bumping 4.14, the kernel build failed due to missing CONFIG_KASAN
symbol. Move it to generic config instead of defining it for all arm64
and x86/64 targets.
It was only added in 4.0, so not needed in config-3.18.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
Remove a stray -Wp left in host_c_flags causing build failures for newer
4.14 versions.
Reported-by: Michael Marley <michael@michaelmarley.com>
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
A few UBI messages lacks the trailing newline character which
leads to ugly lines in the bootlog like this:
[ 6.649159] UBI error: no valid UBI magic found inside mtd6[ 6.667751] Freeing unused kernel memory: 2196K
Add a newline character to the end of the messages to fix it.
After the fix the line from above looks better in the log:
[ 6.609182] UBI error: no valid UBI magic found inside mtd6
[ 6.627599] Freeing unused kernel memory: 2132K
Signed-off-by: Gabor Juhos <juhosg@freemail.hu>
A few UBI messages lacks the trailing newline character which
leads to ugly lines in the bootlog like this:
[ 6.649159] UBI error: no valid UBI magic found inside mtd6[ 6.667751] Freeing unused kernel memory: 2196K
Add a newline character to the end of the messages to fix it.
After the fix the line from above looks better in the log:
[ 6.609182] UBI error: no valid UBI magic found inside mtd6
[ 6.627599] Freeing unused kernel memory: 2132K
Signed-off-by: Gabor Juhos <juhosg@freemail.hu>
A few UBI messages lacks the trailing newline character which
leads to ugly lines in the bootlog like this:
[ 6.649159] UBI error: no valid UBI magic found inside mtd6[ 6.667751] Freeing unused kernel memory: 2196K
Add a newline character to the end of the messages to fix it.
After the fix the line from above looks better in the log:
[ 6.609182] UBI error: no valid UBI magic found inside mtd6
[ 6.627599] Freeing unused kernel memory: 2132K
Signed-off-by: Gabor Juhos <juhosg@freemail.hu>
In kernel 4.14 kmod-bluetooth depends on kmod-crypto-ecdh, add
kmod-crypto-ecdh to LEDE.
Both packages also depend on the kmod-crypto-kpp package. To build this
we have to fix the dependency of CRYPTO_ECDH which has a typo.
This patch is already accepted upstream.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
This adds initial support for kernel 4.14 based on the patches for
kernel 4.9.
In the configuration I deactivated some of the new possible security
features like:
CONFIG_REFCOUNT_FULL
CONFIG_SLAB_FREELIST_HARDENED
CONFIG_SOFTLOCKUP_DETECTOR
CONFIG_WARN_ALL_UNSEEDED_RANDOM
And these overlay FS options are also deactivated:
CONFIG_OVERLAY_FS_INDEX
CONFIG_OVERLAY_FS_REDIRECT_DIR
I activated this:
CONFIG_FORTIFY_SOURCE
CONFIG_POSIX_TIMERS
CONFIG_SLAB_MERGE_DEFAULT
CONFIG_WATCHDOG_HANDLE_BOOT_ENABLED
I am not sure if I did the porting correct for the following patches:
target/linux/generic/backport-4.14/020-backport_netfilter_rtcache.patch
target/linux/generic/hack-4.14/220-gc_sections.patch
target/linux/generic/hack-4.14/321-powerpc_crtsavres_prereq.patch
target/linux/generic/pending-4.14/305-mips_module_reloc.patch
target/linux/generic/pending-4.14/611-netfilter_match_bypass_default_table.patch
target/linux/generic/pending-4.14/680-NET-skip-GRO-for-foreign-MAC-addresses.patch
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
These are taken from the x86 target and should make support kernel 4.9
and 4.14 in the x86 target easier.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
When mtdsplit_minor.c is compiled under Linux 4.4, the compiler
drops the following warning:
CC drivers/mtd/mtdsplit/mtdsplit_minor.o
drivers/mtd/mtdsplit/mtdsplit_minor.c:106:14: warning: initialization from incompatible pointer type [-Wincompatible-pointer-types]
.parse_fn = mtdsplit_parse_minor,
^
drivers/mtd/mtdsplit/mtdsplit_minor.c:106:14: note: (near initialization for 'mtdsplit_minor_parser.parse_fn')
The second parameter of the parser function must not have a 'const'
qualifier in 4.4. The 001-mtdsplit_backport.patch removes the qualifier
from other partition parsers. Update it to handle mtdsplit_minor.c as
well.
Signed-off-by: Gabor Juhos <juhosg@freemail.hu>
Currently local TCP performance on wifi devices can be limited because
the TSQ (TCP Small Queues) code is tuned for wired ethernet latencies.
With this patch drivers can increase the amount of local buffering to
allow TCP to trigger larger aggregation sizes
This commit is modified from the upstream version to allow #ifdef based
backport feature detection
Signed-off-by: Felix Fietkau <nbd@nbd.name>
The splitter ignored the rootfs offset from the header, probably
because until c1e6e61 it was invalid.
This patch fixes the splitter to use the now correct header data.
Regarding target/linux/ar71xx/files/drivers/mtd/tplinkpart.c,
this particular splitter "falls back" to the correct rootfs offset
reading and as such it doesn't need to be updated, although it will
report a kernel partition length that can be larger than the actual
length as it assumes that partition fills the entire segment up to
the rootfs partition.
Tested-by: Mathias Kresin <dev@kresin.me>
Tested-by: Stefan Lippers-Hollmann <s.l-h@gmx.de>
Tested-by: Daniel Engberg <daniel.engberg.lists@pyret.net>
Tested-by: Henryk Heisig <hyniu@o2.pl>
Signed-off-by: Thibaut VARÈNE <hacks@slashdirt.org>
Some targets need 4K sectors for small flash chips (e.g. some
routerboards, where the entire chip is just one "erase block"), whereas
on other devices 4K sectors lead to horrible flash erase/write
performance.
Set the default limit in the generic kernel configuration to 4 MiB to
ensure that all new platforms don't use 4K sectors for bigger flash
chips. On all existing targets use 16 MiB for now to avoid regressions.
They will be changed individually in follow-up commits.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Without this, posix_[fm]advise does not work. This causes issues with
btrfs-progs, which uses fadvise to drop caches.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Patch 610-netfilter_match_bypass_default_check added an extra flag IPT_F_NO_DEF_MATCH
which is copied to user space in function copy_entries_to_user. The 32bit compat
layer function was missing the same logic to copy the flag IPT_F_NO_DEF_MATCH to
user space for a 64bit kernel and 32 bit user space.
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
Signed-off-by: Thierry Dutre <thierry.dutre@dtsystems.be>
Patch 610-netfilter_match_bypass_default_check added an extra flag IPT_F_NO_DEF_MATCH
which is copied to user space in function copy_entries_to_user. The 32bit compat
layer function was missing the same logic to copy the flag IPT_F_NO_DEF_MATCH to
user space for a 64bit kernel and 32 bit user space.
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
Signed-off-by: Thierry Dutre <thierry.dutre@dtsystems.be>
Patch 610-netfilter_match_bypass_default_check added an extra flag IPT_F_NO_DEF_MATCH
which is copied to user space in function copy_entries_to_user. The 32bit compat
layer function was missing the same logic to copy the flag IPT_F_NO_DEF_MATCH to
user space for a 64bit kernel and 32 bit user space.
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
Signed-off-by: Thierry Dutre <thierry.dutre@dtsystems.be>
