Check memblock regions for sufficient size before attempting to use
them. Allow checks for multiple memblock regions until a suitable one is
found.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Our code was assuming CPU port uses the highest number. My BCM53573
device has eth0 connected to port 8 and eth1 connected to port 5. While
working on support for it I tried to:
1) Enable all ports (including port 8)
2) Set CPU port to 5
I noticed port 8 is not accessible anymore. It was just a development
process but it seems like something worth fixing anyway.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Acked-by: Jonas Gorski <jonas.gorski@gmail.com>
With that patch in place for initramfs no additional options are
reported for "/" partition. What's really important is missing
info about sizes. Which in its turn makes opkg think that there's
no space on "/" partition to install software.
I understand that's a sort of corner-case, people rarely install
packages on ramfs but anyways why not?
Just in case that's what I see with the patch:
---------------------->8--------------------
root@lede:/# cat /proc/mounts
rootfs / rootfs rw 0 0
proc /proc proc rw,nosuid,nodev,noexec,noatime 0 0
sysfs /sys sysfs rw,nosuid,nodev,noexec,noatime 0 0
tmpfs /tmp tmpfs rw,nosuid,nodev,noatime 0 0
tmpfs /dev tmpfs rw,nosuid,relatime,size=512k,mode=755 0 0
devpts /dev/pts devpts rw,nosuid,noexec,relatime,mode=600 0 0
debugfs /sys/kernel/debug debugfs rw,noatime 0 0
---------------------->8--------------------
And without:
---------------------->8--------------------
root@lede:/# cat /proc/mounts
rootfs / rootfs rw,size=256168k,nr_inodes=32021 0 0
proc /proc proc rw,nosuid,nodev,noexec,noatime 0 0
sysfs /sys sysfs rw,nosuid,nodev,noexec,noatime 0 0
tmpfs /tmp tmpfs rw,nosuid,nodev,noatime 0 0
tmpfs /dev tmpfs rw,nosuid,relatime,size=512k,mode=755 0 0
devpts /dev/pts devpts rw,nosuid,noexec,relatime,mode=600 0 0
debugfs /sys/kernel/debug debugfs rw,noatime 0 0
---------------------->8--------------------
Note how different is entry for rootfs.
And given there's no known rationale for that patch we're
getting rid of it.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Jonas Gorski <jonas.gorski@gmail.com>
Cc: Rafał Miłecki <zajec5@gmail.com>
Cc: John Crispin <john@phrozen.org>
Cc: Felix Fietkau <nbd@nbd.name>
This is required to update bcma without build breakage. One of bcma
patches changes BCMA_SFLASH dependency.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
On UBI-enabled devices using squashfs as their rootfs an error
message like
UBIFS error (ubi0:3 pid 1): init_constants_early: too few LEBs (12), min. is 17
was thrown while probe-mounting the rootfs which later on succeeds and
thus shouldn't alert the user.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Instead of disabling unwinding entirely this upstream patch
just disables generation of async unwind tables.
Once the patch in question lands in stable 4.4 tree this change
essentially must be removed (otherwise patch application will fail).
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
When USB Wi-Fi dongle based on Atheros AR9271 is connected to OHCI
(USB 1.1) controller following warnings flood debug console:
------------------------>8---------------------------
usb 1-1: new full-speed USB device number 2 using ohci-platform
usb 1-1: ath9k_htc: Firmware ath9k_htc/htc_9271-1.4.0.fw requested
usb 1-1: ath9k_htc: Transferred FW: ath9k_htc/htc_9271-1.4.0.fw, size: 51008
------------[ cut here ]------------
WARNING: CPU: 0 PID: 4 at drivers/usb/core/urb.c:450
usb_submit_urb+0x162/0x404
usb 1-1: BOGUS urb xfer, pipe 1 != type 3
Modules linked in:
CPU: 0 PID: 4 Comm: kworker/0:0 Not tainted 4.6.3 #10
Workqueue: events request_firmware_work_func
Stack Trace:
arc_unwind_core.constprop.1+0x94/0x10c
---[ end trace 2249b79eac9991d1 ]---
------------[ cut here ]------------
WARNING: CPU: 0 PID: 4 at drivers/usb/core/urb.c:450 usb_submit_urb+0x162/0x404
usb 1-1: BOGUS urb xfer, pipe 1 != type 3
Modules linked in:
CPU: 0 PID: 4 Comm: kworker/0:0 Tainted: G W 4.6.3 #10
Workqueue: events request_firmware_work_func
Stack Trace:
arc_unwind_core.constprop.1+0x94/0x10c
---[ end trace 2249b79eac9991d2 ]---
------------[ cut here ]------------
WARNING: CPU: 0 PID: 4 at drivers/usb/core/urb.c:450 usb_submit_urb+0x162/0x404
usb 1-1: BOGUS urb xfer, pipe 1 != type 3
Modules linked in:
CPU: 0 PID: 4 Comm: kworker/0:0 Tainted: G W 4.6.3 #10
Workqueue: events request_firmware_work_func
Stack Trace:
arc_unwind_core.constprop.1+0x94/0x10c
---[ end trace 2249b79eac9991d3 ]---
...
------------------------>8---------------------------
With removed warning Wi-Fi dongle works properly.
Even though this is not the best solution it gets us a working Wireless
AP. Anyways new discussion was started in linux-usb mailing list to find
a proper solution instead of that hack.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
The kernel config option CONFIG_NET_UDP_TUNNEL is not visible and can
not directly be activated. When kmod-udptunnel4 or kmod-udptunnel6 are
build these packages could be empty when no other kernel module selects
CONFIG_NET_UDP_TUNNEL.
Reported-by: Baptiste Jonglez <baptiste@bitsofnetworks.org>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
This patch is a backport for current LEDE 4.4 Kernels.
It is already upstream, for linux-next and stable.
The initial commit message is below:
The bridge is falsly dropping ipv6 mulitcast packets if there is:
1. No ipv6 address assigned on the brigde.
2. No external mld querier present.
3. The internal querier enabled.
When the bridge fails to build mld queries, because it has no
ipv6 address, it slilently returns, but keeps the local querier enabled.
This specific case causes confusing packet loss.
Ipv6 multicast snooping can only work if:
a) An external querier is present
OR
b) The bridge has an ipv6 address an is capable of sending own queries
Otherwise it has to forward/flood the ipv6 multicast traffic,
because snooping cannot work.
This patch fixes the issue by adding a flag to the bridge struct that
indicates that there is currently no ipv6 address assinged to the bridge
and returns a false state for the local querier in
__br_multicast_querier_exists().
Special thanks to Linus Lüssing.
Signed-off-by: Daniel Danzberger <daniel@dd-wrt.com>
This adds a configuration options that allows to make filesystem ACL support
the default in the kernel, except for old nfs.
Signed-off-by: Daniel Dickinson <openwrt@daniel.thecshore.com>
This patch is already included in the Linux mainline kernel since
v3.15, remove it from LEDE, see the lines directly before this patch.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
This patch was introduced in commit r16412 for the brcm47xx target only
and then moved to generic in commit r32395. It was initially added
because of ticket #5186 and should fix some problems with fuse file
systems and MIPS caches. The commit comment in r32395 says that this a
generic problem in MIPS CPUs, but does not name any specifics about
that. There was a fix added to kernel 2.6.21 in commit commit
7575a49f20 "[MIPS] Implement flush_anon_page()." that should fix this
problem, but that was already available before both commits were done
to OpenWrt.
I just tested fuse with ntfs.3g without this patch on a BCM4704
(BMIPS3300 V0.6) SoC and haven't seen any problems. Someone reported
that removing this patch improves some fuse operations by 5 times on
some modern MIPS cores.
My test was only a simple "dd if=/dev/zero of=/mnt/zero bs=5000" to an
USB stick.
This patch removes the patch to OpenWrt, because I assume that it is
not needed any more and Felix, the orginal author, also thinks so.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
In the upstream kernel and the upstream squashfs4 tools the xz
compression header looks the following:
struct disk_comp_opts {
__le32 dictionary_size;
__le32 flags;
};
We added some other members and also moved some existing members. Place
the members which are already in upstream header at the same position
as in that kernel and add our own at the end. The kernel should not
have a problem when there are some additional members and just ignore
them.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
In kernel 4.7 there is upstreamed b53 driver using (mostly?) the same
symbols as our b53 does. Change our symbols so both drivers can coexist
in kernel tree.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Acked-by: Jonas Gorski <jonas.gorski@gmail.com>
So far "kernel" partition didn't contain just a kernel. It also included
Seama header and meta data. This was making kernel update complex and it
wasn't trivial to read kernel size.
Fix it by making "kernel" parition contain just a kernel image.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Despite the MS_SILENT flag being set when probing for ubifs rootfs a
logline indicating an error is generated during boot:
UBIFS error (pid: 1): cannot open "ubi0:rootfs", error -19
This leads to confusion and there shouldn't be lines containing
the word 'error' twice in a bootlog if actually everything is fine
(just the rootfs happens to be something else than ubifs)
The patch added has been submitted and was accepted upstream, see:
http://lists.infradead.org/pipermail/linux-mtd/2016-June/068056.htmlhttp://patchwork.ozlabs.org/patch/637491
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
This can be used to prevent double compression for platforms where the
boot loader already expects compressed images.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
The swconfig kernel infrastructure fails to do any permissions checks when
changing settings. As such an ordinary user account on a device with a
switch can change switch settings without any special permissions.
Routers generally have few non-admin users so this isn't a big hole, but it
is a security hole. Likely the greatest danger is for multifunction devices
which have a lot of extra daemons, compromising a low-security daemon would
allow one to modify switch settings and cause the router/switch to appear to
lock-up (or cause other sorts of troublesome nyetwork behavior).
Implement a check for CAP_NET_ADMIN in swconfig_set_attr() and deny any
requests originating from user contexts lacking this capability.
Reported-by: Elliott Mitchell <ehem+openwrt@m5p.com>
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
OpenWRT changed the default fq_codel sch->limit from 10240 to 1024,
without also adjusting q->flows_cnt. Eric Dumazet explains below that
you must also adjust the buckets (q->flows_cnt) for this not to break.
Eric explains: Limit of 1024 packets and 1024 flows is not wise I think.
(If all buckets are in use, each bucket has a virtual queue of 1 packet,
which is almost the same than having no queue at all)
I suggest to have at least 8 packets per bucket, to let Codel have a
chance to trigger. So you could either reduce number of buckets to 128
(if memory is tight), or increase limit to 8192.
flows_cnt is now set to 1024/8=128
Signed-off-by: Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk>
Building for octeon fails with
'arch/mips/vdso/vdso-n32.so.dbg' already contains a '.MIPS.abiflags'
section
if the file already exists from a prior build.
Use the same workaround as the one for vdso.so.dbg committed in
9eb155353a.
Commit 91f205acaf extended the workaround
to cover vdso-o32.so.dbg but missed the vdso-n32.so.dbg which is added
now by this change.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Building for octeon fails with
'arch/mips/vdso/vdso-o32.so.dbg' already contains a '.MIPS.abiflags'
section
if the file already exists from a prior build.
Use the same workaround as the one for vdso.so.dbg committed in
9eb155353a.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
Currently the build fails with
'arch/mips/vdso/vdso.so.dbg' already contains a '.MIPS.abiflags' section
if the file already exists from a prior build.
Add a makefile rule to force the rebuild of vdso.so.dbg if genvdso has
has been changed to workaround the failure.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
This enables misaligned access handling by software in Linux kernel.
With some wireless drivers (ath9k-htc and mt7601u for example) we see
misaligned accesses here and there and to cope with that without
fixing stuff in the drivers we're just gracefully handling it on ARC.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
SVN-Revision: 49134
The current code only partially invalidates both caches
because the cache size and cache-line size values are
incorrectly passed to the C code.
Fix the assembly code to pass the arguments in the correct
order.
Tested on RB532.
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
SVN-Revision: 49056
The function memblock_insert_region() is in the section
__init_memblock, also put crashlog_init_memblock there.
This fixes this section mismatch warning:
The function memblock_insert_region.isra.1() references
the function __meminit crashlog_init_memblock().
This is often because memblock_insert_region.isra.1 lacks a __meminit
annotation or the annotation of crashlog_init_memblock is wrong.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 48931
The buildbots complained about these config options being missing for arm64:
CONFIG_DUMMY_CONSOLE_COLUMNS=80
CONFIG_DUMMY_CONSOLE_ROWS=25
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 48929
Combine all bus operations for one MMD access in one function.
Protecting all these bus operations with one lock also helps
to avoid potential issues due to bus operations intercepting
the register and data write.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 48914
The default TTL for address resolution table entries is 5 minutes
for all members of the AR8216 family. This can cause issues if
e.g. Wifi clients roam to another AP and their MAC appears on
another switch port suddenly. Then the client may not be reachable
until the old ARL entry expires.
I would have expected the switch to invalidate old entries if it
detects the same MAC on another port. But that's not the case.
Therefore make the TTL for ARL entries configurable.
The effective TTL will always be a multiple of 7 seconds.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 48913
The line before includes the port number anyway so there's no need
to duplicate the port number in the MIB info header.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 48912
The decimal values especially for TxByte and RxGoodByte are hard to read
once bigger amounts of data have been transferred.
Therefore complement the decimal values with info in GiB / MiB / KiB.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 48911
For unused switch ports all MIB values are zero. Displaying ~40 empty
MIB counters is just confusing and makes it hard to read the output of
swconfig dev <dev> show.
Therefore, if all MIB counters for a port are zero, just display
an info that the MIB counters are empty.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 48910
Enable platform-supplied WLAN LED name for ath9k device.
Signed-off-by: Michal Cieslakiewicz <michal.cieslakiewicz@wp.pl>
Acked-by: Hartmut Knaack <knaack.h@gmx.de>
SVN-Revision: 48879
The vdso version of this function has some problems with the cache.
Very often it works on dated data which causes problem. We are
currently working on fixing this in upstream Linux kernel.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 48787
This patch adds speed_mask special file to LEDs connected to switch ports
via 'switch' trigger. It allows to choose which speeds to signal when link
is up. If router has more than one LED per port, they may light up
differently depending on how fast connection is. Default setting is 'all
speeds' so backward compatibility with system scripts (for example uci) is
maintained.
Signed-off-by: Michal Cieslakiewicz <michal.cieslakiewicz@wp.pl>
SVN-Revision: 48775
This patch changes swconfig_trig_port_mask_store() handler to utilize
kstrtoul() function instead of call to obsolete simple_strtoul(). Thanks
to this change, new handler takes less memory and makes port_mask special
file accept not only hexadecimal, but also decimal and octal numbers.
Signed-off-by: Michal Cieslakiewicz <michal.cieslakiewicz@wp.pl>
SVN-Revision: 48774
Seama format has 2 similar headers: container (seal) header and entity
header. The first one has size always set to 0 and doesn't contain MD5
digest.
When dealing with Seama on a flash we deal directly with an entity. You
can see mtdsplit_parse_seama reads from offset 0 and expects entity to
be there. Seama container is used by bootloader / interface only which
extract entity out of it and flash it.
That said we should fix our header struct. This is important as we
calculate possible rootfs offset assuming it may be placed right after
Seama entity. So far calculate offset was always 16B too low.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 48754
When dealing with Broadcom hardware we can simply use swconfig's generic
helper, we just need to do some validation of requested state.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 48623
Thanks to this change swconfig can access port PHYs e.g. when setting
port link state with a generic helper.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 48622
It's quite common for switches to have PHY per port so adding a generic
helper setting link state will help many drivers. It just needs an API
to access PHYs which this patch also adds.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 48621
Some D-Link routers (e.g. DIR-885L) have NAND and use Seama format. It
means OpenWrt will want to have UBI in Sseama entity and should be able
to detect it.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 48600
Our mtdsplit parsers may want to create partition with name choice based
on partition file system (e.g. SquashFS vs. JFFS2). This patch allows
passing extra argument pointing to variable that will be set properly.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 48598
Rename kernel_size variable as it includes whole entity size, not just a
kernel size. Also update comments to match it and describe better what
are we checking/looking for.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 48480
The new .abiflags section was kept, pushing the appended arguments to
the wrong offset and causing it to read the section instead, making
it fail on boot.
Fix this by dropping this section as well as the other sections.
Closes#21679.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 48474
Directly return the return value of genl_register_family_with_ops()
instead of storing it in a temporary variable, then returning it.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 48472
This reverts commit r48335
The workaround is incomplete and cannot cover all possible cases. The
only real solution to this problem is to disable this feature on
ARM11MPcore entirely.
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 48460
BCM531x5 has two pontential cpu ports, and header mode can be enabled
independently on both.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 48302
This add support for IGMP Snooping on atheros switches (disabled by default),
which avoids flooding the network with multicast data.
Tested on TL-WDR4300: disabling IGMP Snooping results in multicast flooding
on each specific port, enabling it back again prevents each port from
receiving all multicast packets.
Partially based on: http://patchwork.ozlabs.org/patch/418122/
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
SVN-Revision: 48268
On most image types the rootfs ends at an erase-block. However, at least
with brnImages this is not the case: while the partitions are aligned
with the erase-block size there is a 12 byte footer at the end of the
partition which must not be touched by any filesystem. This lead to a
rootfs_data partition which was not aligned properly (and thus ended up
being readonly):
0x000000480000-0x00000085a800 : "rootfs_data" (128 KiB EB)
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
SVN-Revision: 48263
This allows splitting EVA images (usually found in fritz devices). The
firmware will be split into a kernel and a separate rootfs partition.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
SVN-Revision: 48262
This adds brnImage (used with the brnboot bootloader) firmware parsing
support. brnboot verifies the integrity of the firmware stored on the
"Code Image" partitions by looking at the 12 byte footer at the very end
of the partition. This footer contains the checksum of the original
brnImage (kernel + rootfs/squashfs) and must not be touched (by our JFFS2
rootfs_data - otherwise the image will not be bootable anymore).
Big thanks to Mathias Kresin for analyzing the brnImage structure and
finding out the information how to keep images valid even when adding a
nested rootfs_data partition.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
SVN-Revision: 48261
Given those patches are relevant to any ARC platform and even
ISA version it makes perfect sense for patches to exist
in one place instead of being duplicated for each new ARC-based ASIC.
Note this is a prerequisite for upstreaming of ARC HS38 support in
OpenWRT.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Felix Fietkau <nbd@openwrt.org>
Cc: Jo-Philipp Wich <jow@openwrt.org>
Cc: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 48241
This removes one patch which was applied upstream with commit
67b9bcd36906e12a15ffec19463afbbd6a41660e. All other patches were
refreshed.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
SVN-Revision: 48203
Some switches can force link speed for a port. Let's add API that will
allow drivers to export this feature.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 48142
This fixes regression introduced in my recent ledtrig-netdev commit.
Events triggered by different interfaces were stopping timer so it
wasn't working for tx/rx mode.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 48072
It was never tested, most likely not working (because of le32_to_cpu)
and not upstreamed.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 48060
The ppp0 interface is renamed after the connection is established. Due
to a missing NETDEV_REGISTER event, the ledtrig-netdev isn't aware of
the renamed interface and literally ignores the device
(no tx/rx indication, led isn't switched off with 'ifdown wan').
Signed-off-by: Mathias Kresin <openwrt@kresin.me>
SVN-Revision: 48048
As explained earlier, using SWITCH_TYPE_LINK gives more flexibility,
it doesn't require e.g. string parsing to read some data.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 47999
So far we were sending link data as a string. It got some drawbacks:
1) Didn't allow writing clean user space apps reading link state. It was
needed to do some screen scraping.
2) Forced whole PORT_LINK communication to be string based. Adding
support for *setting* port link required passing string and parting
it in the kernel space.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 47997
Previously switching to non-existing device (interface) could result in
leaving LED on.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 47990
We may just delete timer on every trigger update and then start it again
if needed. This will let us avoid both: races and locking in frequently
called timer callback.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 47987
Read/write lock was adding useless complexity, there wasn't any real
gain in case of this driver.
Also switch to _bh variants to avoid deadlocks.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 47986
All supported kernels require patching ledtrig-netdev in the same way,
so it's safe to just move these changes to the base version of this
driver. We needed these patches for some old kernels 2.6.36 and 3.11.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 47962
In r45970 the MAC swap handling was made opt-in, however some boards
have been forgotten during the conversion. Since the reference design
uses this MAC swapping, and pretty much all known boards using this chip
seem to do so too, enabling the swapping is a more reasonable default
than leaving it disabled.
Change the code to still allow boards to opt-out of this.
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 47956
This replaces old bcm53xx patch for scanning whole flash and makes
bcm47xxpart compatible with NAND.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 47800
Now that IPV6 isn't build as a module anymore, we don't need to decouple
bridge from ipv6 anymore.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 47751
In old days mtd_add_partition was checking for partitions overlapping
which was breaking our splitting feature. We had to modify this function
by adding an extra bool dup_check parameter. Upstream commit:
3a434f66e6da ("mtd: part: Remove partition overlap checks")
removed that check so we don't need our modification anymore.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 47748
For a long time we have a dedicated firmware splitter for uimage as a
replacement for the one builtin into mtd subsystem. For a year we got a
proper warning for users sticking to the old one (r43700). It should be
really safe to drop it now, especially since we do it for 4.4 only.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 47747
Overlayfs stacking was broken in 4.1.12 but now fixed in upstream branch.
Stacking allows experimenting with configurations without the need to
reset a device to factory config.
Signed-off-by: Roman Yeryomin <roman@advem.lv>
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 47714
Support is present in 3.19 and above as kernel commit:
fcc87a95195236b0935183361a72e4a98bf577d8
Changes to other existing patches are a result of "make target/linux/refresh".
Signed-off-by: Srdjan Rosic <srdjan.rosic@gmail.com>
Signed-off-by: Srdjan Rosic <rosic@google.com>
SVN-Revision: 47588
compat-wireless/backports now contains a bcm47xx_nvram.h file to
backport some of the functions in it which are used by the bcmfmac
driver. This file just checks for the kernel versions and provide an
empty implementations on older kernel versions. This is OK on most
systems, but on bcm47xx / bcm53xx systems we want to call the real
functions here. This commit removes the file from backports in our
build process like we do it with the bcma and ssb header files. Instead
we add a recent version into our kernel so all code uses only one
header file. On bcm47xx / bcm53xx the real implementations of this code
will be used.
Reported-by: Hante Meuleman <meuleman@broadcom.com>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 47467
This fix contains 2 parts:
- kernel 4.1: backport upstream patch "perf build: Do not fail on missing Build file"
- add NO_LZMA=1 to perf MAKE_FLAGS to disable LZMA support
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
SVN-Revision: 47338
Kernel 3.14 added aditional genphy_soft_reset phy reset to phy_init_hw in drivers/net/phy/phy_device.c
Since adm6996 does in driver soft reset and doesn't use BMCR_RESET for soft reset
add dummy soft_reset callback to adm6996 driver, like it is done in ar8216.
This fixes ticket #20147
Signed-off-by: Andrej Vlasic <andrej.vlasic0@gmail.com>
SVN-Revision: 47272
This contains two sets of fixes for the 8139cp driver.
For all kernel versions older than 4.3, we can apply the fixes from the
4.3-rc4 kernel. In particular, these fix the TX timeout recovery which
is causing my Geos to lock up until the hardware watchdog kicks in.
For 4.0 and later kernels, we can also apply the additional improvements
which are going into 4.4 to fix and enable hardware checksum/TSO
offload. Backporting those to older kernels is non-trivial.
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
SVN-Revision: 47220
A year ago I made this old splitter obsolete (see r42828), it's time to
completely remove it now. For all this time we got these warnings:
"Dedicated partitioner didn't create "rootfs_data" partition, please fill a bug report!"
and
"Support for built-in "rootfs_data" splitter will be removed, please use CONFIG_MTD_SPLIT_SQUASHFS_ROOT"
but I'm not aware of any bug reports related to that.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 47040
the lantiq dsl hack to struct atm_trafprm broke the ABI fix this by moving
the variables to the end of the struct
Signed-off-by: John Crispin <blogic@openwrt.org>
SVN-Revision: 46964
As requested here #20432 I've
tried to fix the missing symbols from the specified default kernel
config files.
I ran into this problem trying to build an image for x86_64 target
with a custom kernel configuration.
Signed-off-by: Manganini Mirko <mmanganini@gmail.com>
SVN-Revision: 46879
The previous "link" and "status" functions were non-standard,
and thus less useful for parsing.
Signed-off-by: Claudio Leite <leitec@staticky.com>
SVN-Revision: 46864
commit r30917 ("kernel: bypass all netfilter hooks if the sysctls for that
functionality have been disabled - eliminates the overhead of enabling
CONFIG_BRIDGE_NETFILTER in the kernel config") introduced an optimization
which should reduce/eliminate the overhead for traffic send over bridges on
kernels compiled with CONFIG_BRIDGE_NETFILTER=y. But this optimization
breaks the nf_call_iptables per bridge setting which is more fine grained
than the global sysctl net.bridge.bridge-nf-call-iptables setting.
A test reflecting a real world setup was created to identify if this really
eliminates the overhead and if per-bridge nf_call_iptables could be used in
some setups to increase the throughput. A Qualcomm Atheros QCA9558 based
system with one ethernet and an ath9k wifi 3x3 in HT40 mode was used.
Cables from the AP to the wifi station were used to reduce interference
problems during the tests.
The wlan interface was put in one bridge interface called br-wlan. This
bridge usually contains some more wlan interfaces. The eth0 was put in a
second bridge called br-lan. This usually contains some other privileged
wlan or mesh interfaces. Routing was added between br-lan and br-wlan.
Three kernels were tested:
* (default) OpenWrt kernel for this device
* (brfilter-global) OpenWrt kernel with CONFIG_BRIDGE_NETFILTER=y
* (brfilter-local) OpenWrt kernel with CONFIG_BRIDGE_NETFILTER=y and
without 644-bridge_optimize_netfilter_hooks.patch
The changes to the the netfilter settings of the bridge were done via:
* (brfilter-global) /sbin/sysctl -w net.bridge.bridge-nf-call-iptables=1
* (brfilter-lobal) echo 1 > /sys/class/net/br-lan/bridge/nf_call_iptables
and/or echo 1 > /sys/class/net/br-wan/bridge/nf_call_iptables
A station connected to the wlan0 (AP) interface was used to send traffic to
a PC connected via ethernet. iperf with 3 concurrent transmissions was used
to generate the traffic.
| kernel | br-nf-* global | nf-call* iface | download | upload |
|-----------------|----------------|----------------|----------|----------|
| default | 0 | - | 209 | 268 |
| brfilter-global | 0 | - | 185 | 243 |
| brfilter-local | 0 | - | 187 | 243 |
| brfilter-local | 0 | br-lan | 157 | 226 |
| brfilter-local | 0 | br-lan br-wlan | 139 | 161 |
| brfilter-global | 1 | - | 136 | 162 |
Download/upload results in Mibit/s
It can be seen that the patch doesn't eliminate the overhead. It can also
be seen that the throughput of brfilter-global and brfilter-local with
disabled filtering is the roughly the same. Also the throughput for
brfilter-global and brfilter-local for enabled filtering on all bridges is
roughly the same.
But also the brfilter-local throughput is higher when only br-lan requires
the filtering. This setting would not be possible with
644-bridge_optimize_netfilter_hooks.patch applied and thus can only be
compared with brfilter-global and filtering enabled for all interfaces.
Signed-off-by: Sven Eckelmann <sven@open-mesh.com>
SVN-Revision: 46835
It was corrupted in r38528. The most obvious symptom is repeated messages like this:
Tue Sep 8 08:25:18 2015 kern.warn kernel: [77141.972226] br-lan: received packet on wlan0 with own address as source address
Signed-off-by: Dmitry Ivanov <dima@ubnt.com>
SVN-Revision: 46821
Everything except for blkcipher was already built-in, so make blkcipher
built-in as well.
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 46820
Currently, multicast packets from an STA are sent to any according
multicast listener directly through the bridge multicast-to-unicast
feature. Unfortunately, so far this includes the originating STA, too,
resulting in multicast packets being echo'ed back to the originating STA
if it itself is a multicast listener for that group.
This behaviour breaks IPv6 duplicate address detection: An IPv6 Neighbor
Solicitation for IPv6 Duplicate Address Detection is being echo'ed back,
resulting in the host falsely detecting an address collision, which
makes the node unable to claim an IPv6 address and use IPv6 in general.
Mac80211 unfortunately only prevents the echoes for us for multicast
frames. For the multicast frames cast to a unicast destination we'll
need to take care of excluding the originator ourselves.
Signed-off-by: Linus Lüssing <linus.luessing@c0d3.blue>
SVN-Revision: 46765
Add an upstream fix for /proc/net/route causing missing routes doing
several continued reads from it.
Only 4.1+ is affected.
Closes#20403.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 46726
A call to pskb_may_pull() might reallocate skb->data. Therefore we
should only assign the src-pointer after any potential reallocations.
Signed-off-by: Linus Lüssing <linus.luessing@c0d3.blue>
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 46721
This reverts commit a080e8e1943156168913d0353a2e99d1151102aa.
It did not fix the problem but just hid some symptom. The real issue was
that IGMP/MLD report suppression was not considered for the
multicast-to-unicast feature. A recent netifd which isolates IGMP/MLD
reports between STAs by utilizing AP-isolation and bridge-hairpinning
should have fixed this.
It is perfectly fine to apply multicast-to-unicast to IPv6 Neighbor
Solicitations, too (once that feature is configured correctly).
Signed-off-by: Linus Lüssing <linus.luessing@c0d3.blue>
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 46720
Remove 131-MIPS-export-get_c0_perfcount_int.patch which was already applied
in 4.1.6. This fixes the following build error:
arch/mips/ath79/setup.c:217:77: error: redefinition of '__kstrtab_get_c0_perfcount_int'
arch/mips/ath79/setup.c:211:77: note: previous definition of '__kstrtab_get_c0_perfcount_int' was here
arch/mips/ath79/setup.c:217:350: error: redefinition of '__ksymtab_get_c0_perfcount_int'
arch/mips/ath79/setup.c:211:350: note: previous definition of '__ksymtab_get_c0_perfcount_int' was here
scripts/Makefile.build:258: recipe for target 'arch/mips/ath79/setup.o' failed
Reported-by: swalker
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 46715
Properly treat -ENOSYS as no PHY, else ehci-orion won't work without
generic phy support.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 46711
This also clears any bootloader-set FDB defaults. This had
caused issues creating port-based VLANs when mappings
overlapped previous VLANs. Packets destined to a port
not in the default port group flooded all ports.
Tested on a 88E6171 (Linksys EA4500) and 88E6172 ('1900AC)
Signed-off-by: Claudio Leite <leitec@staticky.com>
Signed-off-by: Imre Kaloz <kaloz@openwrt.org>
SVN-Revision: 46699
no-op since 2.6.35
removed in Kernel 4.1
see https://lwn.net/Articles/380931/
Signed-off-by: Dirk Neukirchen <dirkneukirchen@web.de>
SVN-Revision: 46671
The "linux,part-probe" dts parsing is a pretty neat generic feature.
It has been posted to kernel.org and could easily be reused by all
targets.
This change moves the patch to the 3.18 and 4.1 generic folders, and
makes the feature available to all platforms who may want to use it.
Signed-off-by: Mathieu Olivari <mathieu@codeaurora.org>
Acked-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 46654
In addition to the update this also fixes compile problems with kernel 4.1.
This closes#20323.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 46609
This patch was missing for kernel 4.0 and 4.1 in r46464. I also
replaced the version in OpenWrt with the version which went upstream
into the Linux mainline kernel.
This closes#20193 and #20192
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 46493
REGMAP_MMIO is a hidden symbol, so it cannot be directly selected. If nothing
selects it, it will fail the build with:
ERROR: module '/store/buildbot/slave/slave/rb532/build/build_dir/target-mipsel_mips32_musl-1.1.10/linux-rb532/linux-3.18.19/drivers/base/regmap/regmap-mmio.ko' is missing.
make[2]: *** [/store/buildbot/slave/slave/rb532/build/bin/rb532/packages/base/kmod-regmap_3.18.19-1_rb532.ipk] Error 1
Fix this by adding a description to allow selecting it manually.
Fixes: r46451 ("kmod-regmap: enable MMIO support")
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 46485
NCM specs are not actually mandating a specific position in the frame for
the NDP (Network Datagram Pointer). However, some Huawei devices will
ignore our aggregates if it is not placed after the datagrams it points
to. Add support for doing just this, in a per-device configurable way.
While at it, update NCM subdrivers, disabling this functionality in all of
them, except in huawei_cdc_ncm where it is enabled instead.
We aren't making any distinction between different Huawei NCM devices,
based on what the vendor driver does. Standard NCM devices are left
unaffected: if they are compliant, they should be always usable, still
stay on the safe side.
This change has been tested and working with a Huawei E3131 device (which
works regardless of NDP position) and an E3372 device (which mandates NDP
to be after indexed datagrams).
Signed-off-by: Enrico Mioso <mrkiko.rs@gmail.com>
Signed-off-by: Matti Laakso <malaakso@elisanet.fi>
SVN-Revision: 46464
If a link goes down, don't flush the complete ARL table.
Only flush the entries for the respective port.
Don't touch ARL table if a link goes up.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 46381
Adds functions for flushing ARL table entries per port.
Successfully tested on AR8327. Implementation for AR8216/AR8236/AR8316
is based on the AR8236 datasheet and assumes that the three chips
share a common ATU register layout.
Compile-tested only for AR8216/AR8236/AR8316.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 46380
Adds the chip-specific part of reading ARL table for AR8216/AR8236/AR8316.
It's based on the AR8236 datasheet and compile-tested only as I couldn't
find datasheets for AR8216/AR8316 and don't own devices with these chips.
The existing ar8216_atu_flush implementation was used for all three
chip types, therefore I guess they share a common ATU register layout.
More testing would be appreciated.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 46379
To improve reproducibility, prevent the inclusion of timestamps
in the gzip header.
Signed-off-by: Reiner Herrmann <reiner@reiner-h.de>
SVN-Revision: 46361
The mips reloc patch introduced new allocations which were done before
add_unformed_module but never freed them in case of an error. A new hook in
Linux 3.19 called module_arch_freeing_init can be used for freeing memory
which were allocated during this init phase.
The problem can be seen when trying to load a module (via busybox insmod)
when it was already loaded.
free -m
for i in `seq 1 100`; do
/sbin/insmod /lib/modules/*/ath9k.ko >& /dev/null
done
free -m
This simple loop would leak ~3.2 MB.
Signed-off-by: Sven Eckelmann <sven@open-mesh.com>
SVN-Revision: 46247
Backport for the Spansion S25FL164K
It's a 8 MiB flash chip with 4 KiB erase sectors.
Signed-off-by: L. D. Pinney <ldpinney@gmail.com>
SVN-Revision: 46237
This is useful if the device also has an ethernet WAN interface with a
separate mac address (that is derived from the LAN mac address).
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
SVN-Revision: 46220
Some machtypes were dropped when 4.0 support was added, and the
incomplete patch was taken over to 4.1.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 46187
Code calling fpu_emulator_cop1Handler will pass on fault_addr, making gcc
complain about it not being initialized when the FPU emulator is disabled.
Fixes:
arch/mips/kernel/traps.c: In function 'do_fpe':
arch/mips/kernel/traps.c:864:22: error: 'fault_addr' may be used uninitialized in this function [-Werror=maybe-uninitialized]
process_fpemu_return(sig, fault_addr, fcr31);
^
arch/mips/kernel/traps.c: In function 'do_ri':
arch/mips/kernel/traps.c:806:22: error: 'fault_addr' may be used uninitialized in this function [-Werror=maybe-uninitialized]
process_fpemu_return(sig, fault_addr, fcr31);
^
arch/mips/kernel/traps.c:763:15: note: 'fault_addr' was declared here
void __user *fault_addr;
^
arch/mips/kernel/traps.c: In function 'do_cpu':
arch/mips/kernel/traps.c:1421:28: error: 'fault_addr' may be used uninitialized in this function [-Werror=maybe-uninitialized]
if (!process_fpemu_return(sig, fault_addr, fcr31) && !err)
^
cc1: all warnings being treated as errors
make[7]: *** [arch/mips/kernel/traps.o] Error 1
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 46184
The position of the nvram header file on brcm47xx changed with kernel
version 4.1.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 46170
Make some network uapi headers detect if they are included after
not only glibc but also musl headers.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 46161
Boot tested: http://pastebin.com/L6aAb9xj
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
[jogo:
update to 4.1 final
add patches added since submission
delete patches applied in later rcs
restore commit messages in 220-gc-sections and 304-mips_disable_fpu
fix 050-backport_netfilter_rtcache to match new API
update inlined dma ops with upstream changes
add missing config symbols
enabled CONFIG_MULTIUSER
update kmod defintions for 4.1
]
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 46112
On two tested devices: Netgear R6250 (BCM53011 rev 2) and Luxul XWC-1000
(BCM53011 rev 3) it was possible to use port 7 and eth1 (instead of port
5 and eth0). It seems BCM53011 just like BCM53012 has 8 ports and
usually 3 of them are connected to the SoC.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 46104
AR8337 supports a configuration bit to swap MAC0 and MAC6.
Currently this is set in general if an AR8337 is detected and causes
issues with devices using an AR8334 (internally an AR8337, just
less chip pins).
And it might even cause issues with AR8337-based devices with
different board designs.
Swapping the MAC's however isn't needed for AR8337 in general.
It's just needed in case of certain board designs (affected devices
seem to be based on Atheros reference board AP135/136-010).
Therefore this configuration bit should be moved to platform data.
The patch includes the needed changes to the device initialization
code of affected devices. Hopefully I didn't miss any ..
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
SVN-Revision: 45970
On device reset the sizes for the vlan and port tables were wrongly
calculated based on the pointer size instead of the struct size. This
causes buffer overruns on 64 bit targets, resulting in panics.
Fix this by dereferencing the pointers.
Reported-by: Fedor Konstantinov <blmink@mink.su>
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 45938
At least on my b53 chip, the mask is 3 bits wide, and because
of this some STP states are not set properly and discarded when read.
Maybe for some other chips it makes sense to have just 2 bits width,
but I don't have other versions around to test/validate.
If that's the case then maybe we could add another STP state mask.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 45937
STMMAC_PLATFORM and STMMAC_PCI have been added recently in the kernel,
but show up only when STMMAC driver is enabled. So se'll add it in the
generic config, so the kernel build doesn't stall whenever we enable
this driver.
Signed-off-by: Mathieu Olivari <mathieu@codeaurora.org>
SVN-Revision: 45828
With this patch the following two patches are backported:
* bridge: fix parsing of MLDv2 reports
* bridge: allow setting hash_max + multicast_router if interface is down
The former one is an important fix which got just applied to the net-tree
and is queued for stable. The latter is a patch which is needed to make
the hash_max and multicast_router attributes configurable through
netifd.
Signed-off-by: Linus Lüssing <linus.luessing@c0d3.blue>
SVN-Revision: 45783
Update kernel sources and refresh some patches.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Signed-off-by: Imre Kaloz <kaloz@openwrt.org>
SVN-Revision: 45761
Commit 5168c9a5702648eb690d32ec821647aca80aeba9 introduced a regression
during patch application on the 4.0 kernel. Some of the patched content
doesn't match the actual code, therefore leading to the following error:
Applying patch generic/667-ipv6-Fixed-source-specific-default-route-handling.patch
patching file net/ipv6/ip6_output.c
Hunk #1 FAILED at 886.
1 out of 1 hunk FAILED -- rejects in file net/ipv6/ip6_output.c
patching file net/ipv6/route.c
Hunk #1 succeeded at 2247 (offset 2 lines).
Patch generic/667-ipv6-Fixed-source-specific-default-route-handling.patch does not apply (enforce with -f)
This change just adapts the actual patch to fix what is in kernel 4.0
and make it apply cleanly.
Signed-off-by: Mathieu Olivari <mathieu@codeaurora.org>
SVN-Revision: 45705
This chipset has at least 8 usable ports, e.g. Netgear R8000 has ports
5, 7 and 8 connected to Ethernet interfaces:
vlan1ports=0 1 2 3 5 7 8*
vlan2ports=4 8u
Port 6 seems to be always disabled.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Acked-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 45676
Since the kernel/rootfs split handling was modified 2 years ago by r37283 (
https://dev.openwrt.org/changeset/37283 ) and by the subsequent checkins,
users have seen rather scary mtd errors in the log at every boot. The message
ends "-- forcing read-only", which looks a bit error-like. That error has
been mentioned in some forum threads, when users have noticed this message
instead of some actual error.
[ 2.940000] 0x000000070000-0x000000ff0000 : "firmware"
[ 2.970000] 2 netgear-fw partitions found on MTD device firmware
[ 2.970000] 0x000000070000-0x000000188440 : "kernel"
[ 2.980000] mtd: partition "kernel" must either start or end on erase
block boundary or be smaller than an erase block -- forcing read-only
[ 2.990000] 0x000000188440-0x000000ff0000 : "rootfs"
The patch removes the rather useless warning message.
signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
SVN-Revision: 45669
This option has been added in kernel 3.17. It shows-up only when both
ARCH_QCOM and CRYPTO are enabled. So we'll disable these two by default
to avoid stalling the build when these conditions are met.
Signed-off-by: Mathieu Olivari <mathieu@codeaurora.org>
SVN-Revision: 45658
