openwrt/svn-archive/archive.git
8 years agomac80211: brcmfmac: return -ENFILE if interface exists
Rafał Miłecki [Mon, 30 May 2016 06:29:43 +0000 (06:29 +0000)]
mac80211: brcmfmac: return -ENFILE if interface exists

This makes brcmfmac compatible with mac80211 which uses dev_alloc_name
(and so returns -ENFILE on error).

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 49378

8 years agoopenssl: Update to version 1.0.2h
Luka Perkov [Tue, 10 May 2016 22:42:36 +0000 (22:42 +0000)]
openssl: Update to version 1.0.2h

openssl: Update to version 1.0.2h

Bump to the latest version, fixes several security issues:
 * CVE-2016-2107, CVE-2016-2105, CVE-2016-2106, CVE-2016-2109, CVE-2016-2176
More details at https://www.openssl.org/news/openssl-1.0.2-notes.html

Signed-off-by: Mohammed Berdai <mohammed.berdai@gmail.com>
SVN-Revision: 49363

8 years agomtd: seama: exit with error if Seama header wasn't found
Rafał Miłecki [Mon, 9 May 2016 15:34:13 +0000 (15:34 +0000)]
mtd: seama: exit with error if Seama header wasn't found

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r49314

SVN-Revision: 49315

8 years agomtd: seama: fix image data handling
Rafał Miłecki [Mon, 9 May 2016 12:54:53 +0000 (12:54 +0000)]
mtd: seama: fix image data handling

1) Put sanity checks in one place
2) Respect provided offset
3) Read only as much data as needed for MD5 calculation

Thanks to the last change this is a great speedup and memory saver. On
devices with NAND flash we were allocating & reading about 128 MiB while
something about 8 MiB is enough.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r49309

SVN-Revision: 49313

8 years agomtd: seama: move buf allocation to the MD5 function
Rafał Miłecki [Mon, 9 May 2016 12:54:49 +0000 (12:54 +0000)]
mtd: seama: move buf allocation to the MD5 function

This buf is only used in this function now, so lets move it there.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r49308

SVN-Revision: 49312

8 years agomtd: seama: update MD5 using header in the first block buffer
Rafał Miłecki [Mon, 9 May 2016 12:54:46 +0000 (12:54 +0000)]
mtd: seama: update MD5 using header in the first block buffer

This will allow separating first block buffer from a buffer used for MD5
calculation.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r49307

SVN-Revision: 49311

8 years agomtd: seama: add md5 to header struct
Rafał Miłecki [Mon, 9 May 2016 12:54:42 +0000 (12:54 +0000)]
mtd: seama: add md5 to header struct

This allows us to drop some extra offset calculations and simplifies
code a bit.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r49306

SVN-Revision: 49310

8 years agomtd: check for Seama magic early when fixing MD5
Rafał Miłecki [Mon, 9 May 2016 09:55:35 +0000 (09:55 +0000)]
mtd: check for Seama magic early when fixing MD5

This avoid long (and unneeded) process of reading all data in case of
running on MTD not containig Seama entity.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r49304

SVN-Revision: 49305

8 years agomtd: add missing breaks in a switch
Rafał Miłecki [Mon, 9 May 2016 07:55:18 +0000 (07:55 +0000)]
mtd: add missing breaks in a switch

On platforms supporting both: TRX and Seama calling "fixtrx" was
resulting in trying to fix Seama as well.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r49301

SVN-Revision: 49303

8 years agomtd: use tabs for indents
Rafał Miłecki [Mon, 9 May 2016 07:55:08 +0000 (07:55 +0000)]
mtd: use tabs for indents

This makes code style consistent across the whole file.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r49300

SVN-Revision: 49302

8 years agohostapd: backport fix for AP SME drivers with monitor mode support
Rafał Miłecki [Sun, 8 May 2016 21:09:23 +0000 (21:09 +0000)]
hostapd: backport fix for AP SME drivers with monitor mode support

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 49299

8 years agokernel: use backported versions of bgmac patches
Rafał Miłecki [Sun, 8 May 2016 21:09:18 +0000 (21:09 +0000)]
kernel: use backported versions of bgmac patches

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 49298

8 years agobcm53xx: use backported versions of bcma USB patches
Rafał Miłecki [Sun, 8 May 2016 21:09:12 +0000 (21:09 +0000)]
bcm53xx: use backported versions of bcma USB patches

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 49297

8 years agoCC: ubox: turn logd into a separate package
Luka Perkov [Tue, 3 May 2016 12:18:16 +0000 (12:18 +0000)]
CC: ubox: turn logd into a separate package
Currently system log is always included as a part of ubox.
Add logd as a seperate package and add it to default packages list.

Signed-off-by: Andrej Vlasic <andrej.vlasic@sartura.hr>
SVN-Revision: 49294

8 years agomac80211: add support for BCM4366C0 in brcmfmac
Rafał Miłecki [Wed, 27 Apr 2016 07:16:17 +0000 (07:16 +0000)]
mac80211: add support for BCM4366C0 in brcmfmac

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 49261

8 years agoRevert "ar71xx: change some TP-link modelnames (WR841, WA701, WA730)"
John Crispin [Tue, 26 Apr 2016 07:52:01 +0000 (07:52 +0000)]
Revert "ar71xx: change some TP-link modelnames (WR841, WA701, WA730)"

This reverts commit 0fd48fd7dab066db8a46e51775579c9ef507ca25.

We are facing some problem on rolling out the current 15.05-branch to
our devices because the image name changed.

The changed image name breaks compatibility for derived projects and
that's something which should only happen if there is a really good
reason (e.g. security fix).

Signed-off-by: Tim Niemeyer <tim@tn-x.org>
SVN-Revision: 49231

8 years agolantiq: Fix LED handling.
John Crispin [Tue, 26 Apr 2016 07:52:00 +0000 (07:52 +0000)]
lantiq: Fix LED handling.

Backport of r47914 to CC.

Without this patch the TDW8970 is left without any status LED at all,
even during preinit and failsafe.

Signed-off-by: Vittorio Gambaletta <openwrt@vittgam.net>
SVN-Revision: 49230

8 years agolantiq: Fix ath9k LED for TDW8970.
John Crispin [Tue, 26 Apr 2016 07:51:58 +0000 (07:51 +0000)]
lantiq: Fix ath9k LED for TDW8970.

Backport of r47911, r47912, r47969 to CC.

All other required patches have already been backported in r48822.

Signed-off-by: Vittorio Gambaletta <openwrt@vittgam.net>
SVN-Revision: 49229

8 years agoar71x: Added support for TP-Link TL-WR841N v11 to 15.05
John Crispin [Mon, 25 Apr 2016 20:08:54 +0000 (20:08 +0000)]
ar71x: Added support for TP-Link TL-WR841N v11 to 15.05

This patch adds support for the TP-Link TL-WR841N to OpenWrt Chaos Calmer
15.05. Confirmed working [1] and [2].

Signed-off by Stijn Segers <francesco.borromini@inventati.org>

[1] https://forum.openwrt.org/viewtopic.php?id=63657
[2] https://forum.openwrt.org/viewtopic.php?id=61309

SVN-Revision: 49228

8 years agoar71xx: backport support for Netgear WPN824N
John Crispin [Mon, 25 Apr 2016 20:08:52 +0000 (20:08 +0000)]
ar71xx: backport support for Netgear WPN824N

Add support for Netgear WPN824N.
Hardware specs:
  * AR7240, 4 LAN ports, 1 WAN port
  * AR9285 WLAN
  * 32 MB RAM
  * 4 MB Flash
  * 16 LEDs (LAN, WAN and Power/Status contain two LEDs for dual color
    effect)
  * 3 Buttons (not supported)

Signed-off-by: Hartmut Knaack <knaack.h@gmx.de>
SVN-Revision: 49227

8 years agoar71xx: Add TP-Link TL-WA901ND v4 support to Chaos Calmer 15.05.1
John Crispin [Mon, 25 Apr 2016 20:08:49 +0000 (20:08 +0000)]
ar71xx: Add TP-Link TL-WA901ND v4 support to Chaos Calmer 15.05.1

This patch backports support for the TP-Link TL-WA901ND v4 to Chaos Calmer 15.05.1,
which was added into trunk recently [1].

Signed-off by Stijn Segers <francesco.borromini@inventati.org>

[1]: http://patchwork.ozlabs.org/patch/606780/

SVN-Revision: 49226

8 years agoar71xx: Add TP-Link TL-WR740N v6 support to 15.05.1
John Crispin [Mon, 25 Apr 2016 20:08:44 +0000 (20:08 +0000)]
ar71xx: Add TP-Link TL-WR740N v6 support to 15.05.1

This patch adds support for the TP-Link TL-WR740N v6 to OpenWrt Chaos Calmer. Hardware wise, it's very similar to the TL-WR841N v10. See the forum thread [1]. Credit goes to the user 'traveler' who did the legwork and tested builds.

Also thanks to Matthias Schiffer for clarifying the DEVICE_PROFILE settings.

Signed-off by: Stijn Segers <francesco.borromini@inventati.org>

[1]: https://forum.openwrt.org/viewtopic.php?pid=318704

SVN-Revision: 49225

8 years agoCC: ramips: backport rt3352 usbphy fix in dts-file
John Crispin [Mon, 25 Apr 2016 20:08:41 +0000 (20:08 +0000)]
CC: ramips: backport rt3352 usbphy fix in dts-file
Backport of trunk patch. Fix kernel hang during loading ehci-hcd module.

SIgned-Off-By: Serge Vasilugin <vasilugin@yandex.ru>
SVN-Revision: 49224

8 years agoCC: kernel: Support for new Archer C7 with gd25q128 chip
Hauke Mehrtens [Sun, 24 Apr 2016 19:20:10 +0000 (19:20 +0000)]
CC: kernel: Support for new Archer C7 with gd25q128 chip
Archer C7 V2.0 units from December 2015 onwards ( serials starting 215C
) have changed flash chips to the gd25q128 chip, this is supported in
trunk but not presently in 15.05. I would like stable support for this
version so I've back ported the required fix from trunk and removed a
conflicting patch from bcm53xx I've tested the patch it and I'm able to
install a build of 15.05.1 on a new Archer C7 device with this patch.

Signed-off-by: John Marrett <johnf@zioncluster.ca>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 49220

8 years agobcm53xx: bump maximum TRX size for devices with serial flash
Rafał Miłecki [Wed, 20 Apr 2016 18:24:35 +0000 (18:24 +0000)]
bcm53xx: bump maximum TRX size for devices with serial flash

It was already increased for NAND devices so do the same for the ones
with serial memory.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r49207

SVN-Revision: 49208

8 years agoramips: fix 16 bit IO on newer cores
John Crispin [Wed, 20 Apr 2016 16:49:19 +0000 (16:49 +0000)]
ramips: fix 16 bit IO on newer cores

Signed-off-by: John Crispin <blogic@openwrt.org>
SVN-Revision: 49203

8 years agoar71xx: fix nondeterministic hangs during bootconsole/console handover
Rafał Miłecki [Wed, 20 Apr 2016 06:05:56 +0000 (06:05 +0000)]
ar71xx: fix nondeterministic hangs during bootconsole/console handover

Reconfiguring the UART when the FIFO is not empty may cause the boot to
hang. This hang is extremely suspectible to timing differences; recompiling
the kernel with the same configuration, but different UTS_VERSION
timestamps can yield images that hang more or less often. Sometimes images
are produced that hang reproducibly.

This patch should also make it into one of the next linux-stable releases,
but it's better to get this fixed as soon as possible.

Fixes #21773, #21857.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Backport of r49156

SVN-Revision: 49202

8 years agoralink: add spi fix
John Crispin [Tue, 19 Apr 2016 21:01:34 +0000 (21:01 +0000)]
ralink: add spi fix

the fullduplex on CS1 is broken. remove the fullduplex support and run on
plain half duplex on both CS lines.

Signed-off-by: John Crispin <blogic@openwrt.org>
SVN-Revision: 49201

8 years agoralink: add i2c fix
John Crispin [Tue, 19 Apr 2016 21:01:30 +0000 (21:01 +0000)]
ralink: add i2c fix

there is a deadlock that triggers under heavy bus load. until the bug is
fxed, revert to an old version of the I2C driver that does not suffer
from this issue.

Signed-off-by: John Crispin <blogic@openwrt.org>
SVN-Revision: 49200

8 years agoCC: samba: fix some security problems
Hauke Mehrtens [Sun, 17 Apr 2016 11:05:54 +0000 (11:05 +0000)]
CC: samba: fix some security problems
Backport of r49175.

This fixes the following security problems:
* CVE-2015-7560
* CVE-2015-5370
* CVE-2016-2110
* CVE-2016-2111
* CVE-2016-2112
* CVE-2016-2115
* CVE-2016-2118

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 49177

8 years agomac80211: backport brcmfmac workaround for missing boardrev
Rafał Miłecki [Sat, 16 Apr 2016 08:47:12 +0000 (08:47 +0000)]
mac80211: backport brcmfmac workaround for missing boardrev

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 49173

8 years agobcm53xx: backport upstream profiling and USB VCC patches
Rafał Miłecki [Fri, 15 Apr 2016 11:25:27 +0000 (11:25 +0000)]
bcm53xx: backport upstream profiling and USB VCC patches

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 49172

8 years agobcm53xx: backport patch enabling UART on tested devices
Rafał Miłecki [Fri, 15 Apr 2016 11:25:14 +0000 (11:25 +0000)]
bcm53xx: backport patch enabling UART on tested devices

Also rename and simpify patch enabling UART on untested ones.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 49171

8 years agobcm53xx: fix MD5 in Seama format on flash after installing OpenWrt
Rafał Miłecki [Fri, 15 Apr 2016 11:24:57 +0000 (11:24 +0000)]
bcm53xx: fix MD5 in Seama format on flash after installing OpenWrt

We need to fix Seama just like we fix TRX due to the rootfs mark.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r48522

SVN-Revision: 49170

8 years agomtd: enable "fixseama" on bcm53xx
Rafał Miłecki [Fri, 15 Apr 2016 11:24:51 +0000 (11:24 +0000)]
mtd: enable "fixseama" on bcm53xx

There are D-Link bcm53xx devices using Seama format.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r48521

SVN-Revision: 49169

8 years agorpcd: update to latest version
Luka Perkov [Wed, 13 Apr 2016 07:55:22 +0000 (07:55 +0000)]
rpcd: update to latest version

Signed-off-by: Luka Perkov <luka@openwrt.org>
SVN-Revision: 49163

8 years agobcm53xx: fixup 14e4:d612 device class to bridge as well
Rafał Miłecki [Tue, 5 Apr 2016 06:47:11 +0000 (06:47 +0000)]
bcm53xx: fixup 14e4:d612 device class to bridge as well

14e4:d612 was found in D-Link DIR-885L and Netgear R8500. It's most
likely a bridge used by all 14e4:4365 BCM4366 devices.

Without this fixup device wasn't treated as a bridge:
[    2.849895] pci 0000:00:00.0: ignoring class 0x020000 (doesn't match header type 01)
and didn't get its resources assigned. This resulted in child devices
not being able to get any resources (due to the parrent missing them):
[    3.375428] pci 0000:01:00.0: BAR 2: no space for [mem size 0x00400000 64bit]
[    3.387949] pci 0000:01:00.0: BAR 2: failed to assign [mem size 0x00400000 64bit]
[    3.725797] pci 0000:01:00.0: BAR 0: no space for [mem size 0x00008000 64bit]
[    3.738314] pci 0000:01:00.0: BAR 0: failed to assign [mem size 0x00008000 64bit]

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 49118

8 years agobcm53xx: prepare for building D-Link DIR-885L image
Rafał Miłecki [Mon, 4 Apr 2016 14:53:25 +0000 (14:53 +0000)]
bcm53xx: prepare for building D-Link DIR-885L image

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Partial backport of r48606

SVN-Revision: 49117

8 years agofirmware-utils: add oseama tool for creating Seama entities
Rafał Miłecki [Mon, 4 Apr 2016 14:53:22 +0000 (14:53 +0000)]
firmware-utils: add oseama tool for creating Seama entities

It has an important feature (compared to seama) of using multiple input
files, aligning them and padding zeroes until reaching a specified
absolute offset. This is needed for a proper flash layout on NAND. We
want kernel partition to be big enough to handle future updates without
a need to resize it and wipe whole "ubi" partition. It's important as
we don't want to lose block counters.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r48601

SVN-Revision: 49116

8 years agobcm53xx: update BCM5301X patches
Rafał Miłecki [Mon, 4 Apr 2016 14:53:15 +0000 (14:53 +0000)]
bcm53xx: update BCM5301X patches

This moves few upstreamed patches and adds DIR-885L dts file. Please
note it doesn't mean this device is supported yet.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 49115

8 years agokernel: update kernel 3.18 to version 3.18.29
Rafał Miłecki [Wed, 30 Mar 2016 16:43:25 +0000 (16:43 +0000)]
kernel: update kernel 3.18 to version 3.18.29

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r49096

SVN-Revision: 49097

8 years agobcm53xx: improve USB 2.0 PHY support for BCM4709 and BCM47094
Rafał Miłecki [Wed, 30 Mar 2016 12:55:39 +0000 (12:55 +0000)]
bcm53xx: improve USB 2.0 PHY support for BCM4709 and BCM47094

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 49095

8 years agobcm53xx: support USB 3.0 PHY on BCM4707 rev. 6 and BCM47094
Rafał Miłecki [Wed, 30 Mar 2016 10:49:29 +0000 (10:49 +0000)]
bcm53xx: support USB 3.0 PHY on BCM4707 rev. 6 and BCM47094

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 49093

8 years agobcm53xx: simpify USB 3.0 PHY initialization
Rafał Miłecki [Wed, 30 Mar 2016 10:49:25 +0000 (10:49 +0000)]
bcm53xx: simpify USB 3.0 PHY initialization

We can use one code path for BCM4707 rev. 4 and BCM53018.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 49092

8 years agoar71xx: cosmetic fix of alfa ap120c/ap96 ordering
Luka Perkov [Thu, 24 Mar 2016 21:33:21 +0000 (21:33 +0000)]
ar71xx: cosmetic fix of alfa ap120c/ap96 ordering

Signed-off-by: Luka Perkov <luka@openwrt.org>
SVN-Revision: 49087

8 years agouboot-envtools: add support for ap120c
Luka Perkov [Wed, 23 Mar 2016 22:37:05 +0000 (22:37 +0000)]
uboot-envtools: add support for ap120c

Signed-off-by: Luka Perkov <luka@openwrt.org>
SVN-Revision: 49085

8 years agoar71xx: add support for ap120c
Luka Perkov [Wed, 23 Mar 2016 22:37:02 +0000 (22:37 +0000)]
ar71xx: add support for ap120c

Signed-off-by: Luka Perkov <luka@openwrt.org>
SVN-Revision: 49084

8 years agoCC: version.mk and image-config.in: adjust to 15.05.1
John Crispin [Sun, 20 Mar 2016 14:45:02 +0000 (14:45 +0000)]
CC: version.mk and image-config.in: adjust to 15.05.1
* Adjust the default version number in include/version.mk to 15.05.1
* Copy the correct download repo location from include/version.mk
  to base-files/image-config.in

After the change to version.mk, new builds made from CC sources will
have opkg config that downloads from the recent 15.05.1 packages repo
instead of the ancient 15.05 packages repo.

The change to image-config.in ensures that if somebody uses
VERSIONOPT config options in .config, he will get the
correct download repo address (instead of trunk snapshots).

Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
SVN-Revision: 49053

8 years agoar71xx: change some TP-link modelnames (WR841, WA701, WA730)
John Crispin [Sun, 20 Mar 2016 14:44:59 +0000 (14:44 +0000)]
ar71xx: change some TP-link modelnames (WR841, WA701, WA730)

This patch changes the image-name of some TPlink-routers.
Esp. for the TPlink WR841N + WR841ND some users are confused by the fact
that some images have "nd"-names and others "n"-names only.
In addition this also corrects a missed "D" on WA701 and a "-" for
WA730RE.

backport to CC of
r48767 'http://git.openwrt.org/?p=openwrt.git;a=commit;h=dbd0ab83566291e3b4974f4d4289bb7c93b091d6'
r48768 'http://git.openwrt.org/?p=openwrt.git;a=commit;h=656a17c77d530b14355e8fef4a83265ee2c3271d'
r48769 'http://git.openwrt.org/?p=openwrt.git;a=commit;h=783756e3096a417565d58d78e082a29db10738d4'

Signed-off-by: Sven Roederer <devel-sven@geroedel.de>
SVN-Revision: 49052

8 years agoCC: linux-firmware: fix rt73-usb-firmware package
Hauke Mehrtens [Sat, 19 Mar 2016 11:01:06 +0000 (11:01 +0000)]
CC: linux-firmware: fix rt73-usb-firmware package
Backport of r49037

Some parts of this package were named rt73-pci-firmware before which
looks like a copy and past error. This renames all parts to rt73-usb-
firmware and then the firmware gets build and the dependencies from the
package with the kernel module are also working correctly.

This fixes #22069

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 49038

8 years agobrcm47xx: image: create standard TRX images using new building system
Rafał Miłecki [Fri, 11 Mar 2016 21:41:20 +0000 (21:41 +0000)]
brcm47xx: image: create standard TRX images using new building system

Apart from using our new building system there are 2 more changes:

1) Limit amount of images
So far we were generating all standard images (optimized one and two
with no loader) for every SUBTARGET. This is not needed, as e.g. the
only device requiring gzipped kernel is legacy Huawei E970.

2) Change output names
The new image building system requires specifying device name. This
forced picking some and resulted in:
openwrt-brcm47xx-$(SUBTARGET)-squashfs.trx
openwrt-brcm47xx-$(SUBTARGET)-squashfs-gz.trx
openwrt-brcm47xx-$(SUBTARGET)-squashfs-noloader-nodictionary.trx
becoming:
openwrt-brcm47xx-$(SUBTARGET)-standard-squashfs.trx
openwrt-brcm47xx-$(SUBTARGET)-standard-noloader-gz-squashfs.trx
openwrt-brcm47xx-$(SUBTARGET)-standard-noloader-nodictionarylzma-squashfs.trx

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r49006

SVN-Revision: 49007

8 years agomac80211: backport brcmfmac fixes for WARNING and BCM4360 init
Rafał Miłecki [Fri, 11 Mar 2016 09:55:00 +0000 (09:55 +0000)]
mac80211: backport brcmfmac fixes for WARNING and BCM4360 init

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
SVN-Revision: 49004

8 years agobrcm47xx: image: move disabled devices to the new building system
Rafał Miłecki [Fri, 11 Mar 2016 08:54:19 +0000 (08:54 +0000)]
brcm47xx: image: move disabled devices to the new building system

This only drops WGR614 V9 which has 2 MiB flash and it's unlikely to get
any interest.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r48975

SVN-Revision: 49003

8 years agobrcm47xx: image: switch Linksys WRT54G3GV2-VF to the new building system
Rafał Miłecki [Fri, 11 Mar 2016 08:54:16 +0000 (08:54 +0000)]
brcm47xx: image: switch Linksys WRT54G3GV2-VF to the new building system

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r48974

SVN-Revision: 49002

8 years agobrcm47xx: image: switch Netgear WGT634U to the new building system
Rafał Miłecki [Fri, 11 Mar 2016 08:54:13 +0000 (08:54 +0000)]
brcm47xx: image: switch Netgear WGT634U to the new building system

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r48973

SVN-Revision: 49001

8 years agobrcm47xx: image: switch rest of 64k block devices to the new building system
Rafał Miłecki [Fri, 11 Mar 2016 08:54:09 +0000 (08:54 +0000)]
brcm47xx: image: switch rest of 64k block devices to the new building system

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r48972

SVN-Revision: 49000

8 years agobrcm47xx: image: switch Motorola devices to the new building system
Rafał Miłecki [Fri, 11 Mar 2016 08:54:06 +0000 (08:54 +0000)]
brcm47xx: image: switch Motorola devices to the new building system

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r48971

SVN-Revision: 48999

8 years agobrcm47xx: image: switch D-Link DWL-3150 to the new building system
Rafał Miłecki [Fri, 11 Mar 2016 08:54:03 +0000 (08:54 +0000)]
brcm47xx: image: switch D-Link DWL-3150 to the new building system

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r48970

SVN-Revision: 48998

8 years agokernel: rework bcm47xx backport, fix build errors on the brcm47xx target (#21974)
Felix Fietkau [Sat, 5 Mar 2016 10:24:00 +0000 (10:24 +0000)]
kernel: rework bcm47xx backport, fix build errors on the brcm47xx target (#21974)

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 48925

8 years agoCC: curl: detect the library as renamed wolfssl
Hauke Mehrtens [Thu, 3 Mar 2016 22:30:22 +0000 (22:30 +0000)]
CC: curl: detect the library as renamed wolfssl
Without this patch curl does not detect the new cyassl/wolfssl version
and would be build without ssl support instead when cyassl was selected.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 48900

8 years agomac80211: improve rate control performance
Felix Fietkau [Thu, 3 Mar 2016 22:29:13 +0000 (22:29 +0000)]
mac80211: improve rate control performance

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Backport of r48897

SVN-Revision: 48899

8 years agomt76: update to the latest version, adds stability fixes
Felix Fietkau [Thu, 3 Mar 2016 22:29:11 +0000 (22:29 +0000)]
mt76: update to the latest version, adds stability fixes

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Backport of r48895

SVN-Revision: 48898

8 years agoCC: kernel: update kernel 3.18 to version 3.18.27
Hauke Mehrtens [Thu, 3 Mar 2016 22:28:12 +0000 (22:28 +0000)]
CC: kernel: update kernel 3.18 to version 3.18.27
Changelog:
* https://cdn.kernel.org/pub/linux/kernel/v3.x/ChangeLog-3.18.24
* https://cdn.kernel.org/pub/linux/kernel/v3.x/ChangeLog-3.18.25
* https://cdn.kernel.org/pub/linux/kernel/v3.x/ChangeLog-3.18.26
* https://cdn.kernel.org/pub/linux/kernel/v3.x/ChangeLog-3.18.27

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 48896

8 years agoramips: EX2700: remove unused kernel modules
John Crispin [Thu, 3 Mar 2016 20:16:23 +0000 (20:16 +0000)]
ramips: EX2700: remove unused kernel modules

Remove the kmod-mt76 package from the EX2700 profile to free up
around 100 KiB of space on the root filesystem.

This patch applies to both trunk and Chaos Calmer.

Signed-off-by: Joseph C. Lehner <joseph.c.lehner@gmail.com>
SVN-Revision: 48887

8 years agomac80211: add a number of pending fixes
Felix Fietkau [Wed, 2 Mar 2016 14:57:49 +0000 (14:57 +0000)]
mac80211: add a number of pending fixes

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Backport of r48883

SVN-Revision: 48884

8 years agomac80211/hostapd: rework 802.11w driver support selection, do not hardcode drivers...
Felix Fietkau [Wed, 2 Mar 2016 13:06:57 +0000 (13:06 +0000)]
mac80211/hostapd: rework 802.11w driver support selection, do not hardcode drivers in hostapd makefile

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Backport of r46903

SVN-Revision: 48878

8 years agoCC: cyassl: disable Intel ASM for now
Jo-Philipp Wich [Wed, 2 Mar 2016 10:01:48 +0000 (10:01 +0000)]
CC: cyassl: disable Intel ASM for now
With ASM support enabled, CyaSSL fails to build on all x86 subtargets.

Signed-off-by: Jo-Philipp Wich <jow@openwrt.org>
Backport of r48876

SVN-Revision: 48877

8 years agocurl: bump pkg revision
Jo-Philipp Wich [Wed, 2 Mar 2016 09:51:47 +0000 (09:51 +0000)]
curl: bump pkg revision

Signed-off-by: Jo-Philipp Wich <jow@openwrt.org>
SVN-Revision: 48875

8 years agoCC: cyassl: update to wolfssl version 3.8.0 to fix security problems
Hauke Mehrtens [Tue, 1 Mar 2016 22:50:29 +0000 (22:50 +0000)]
CC: cyassl: update to wolfssl version 3.8.0 to fix security problems
Backport of:
r46167: cyassl: version bump to 3.4.6
r46168: cyassl: update to wolfssl 3.6.0
r46551: cyassl: the upstream package in version 4.6.0 changed
r47791: cyassl: update to wolfSSL version 3.7.0
    This version and version 3.6.8 are fixing the following security problems:
    * CVE-2015-7744
    * CVE-2015-6925
r48616: cyassl: update to wolfssl version 3.8.0

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 48873

8 years agoCC: curl: fix CVE-2016-0755
Hauke Mehrtens [Tue, 1 Mar 2016 22:42:51 +0000 (22:42 +0000)]
CC: curl: fix CVE-2016-0755
This fixes the following security problem:
CVE-2016-0755: NTLM credentials not-checked for proxy connection re-use
http://curl.haxx.se/docs/adv_20160127B.html

backport of r48614.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 48872

8 years agoCC: bcm53xx: make patches apply again
Hauke Mehrtens [Tue, 1 Mar 2016 22:11:12 +0000 (22:11 +0000)]
CC: bcm53xx: make patches apply again
Commit r48849 changed the drivers/mtd/spi-nor/spi-nor.c file and broke
this patch in bcm53xx.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 48871

8 years agoiw: update to version 4.3, sync with trunk r47782
Felix Fietkau [Tue, 1 Mar 2016 21:32:22 +0000 (21:32 +0000)]
iw: update to version 4.3, sync with trunk r47782

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 48870

8 years agoCC: openssl: update to 1.0.2g (8 CVEs)
Jo-Philipp Wich [Tue, 1 Mar 2016 15:18:24 +0000 (15:18 +0000)]
CC: openssl: update to 1.0.2g (8 CVEs)
CVE-2016-0704

s2_srvr.c overwrite the wrong bytes in the master-key when applying
Bleichenbacher protection for export cipher suites. This provides a
Bleichenbacher oracle, and could potentially allow more efficient variants of
the DROWN attack.

CVE-2016-0703

s2_srvr.c did not enforce that clear-key-length is 0 for non-export ciphers.
If clear-key bytes are present for these ciphers, they *displace* encrypted-key
bytes. This leads to an efficient divide-and-conquer key recovery attack: if
an eavesdropper has intercepted an SSLv2 handshake, they can use the server as
an oracle to determine the SSLv2 master-key, using only 16 connections to the
server and negligible computation. More importantly, this leads to a more
efficient version of DROWN that is effective against non-export ciphersuites,
and requires no significant computation.

CVE-2016-0702

A side-channel attack was found which makes use of cache-bank conflicts on
the Intel Sandy-Bridge microarchitecture which could lead to the recovery of
RSA keys. The ability to exploit this issue is limited as it relies on an
attacker who has control of code in a thread running on the same hyper-
threaded core as the victim thread which is performing decryptions.

CVE-2016-0799

The internal |fmtstr| function used in processing a "%s" format string in
the BIO_*printf functions could overflow while calculating the length of a
string and cause an OOB read when printing very long strings. Additionally
the internal |doapr_outch| function can attempt to write to an OOB memory
location (at an offset from the NULL pointer) in the event of a memory
allocation failure. In 1.0.2 and below this could be caused where the size
of a buffer to be allocated is greater than INT_MAX. E.g. this could be in
processing a very long "%s" format string. Memory leaks can also occur.
The first issue may mask the second issue dependent on compiler behaviour.
These problems could enable attacks where large amounts of untrusted data is
passed to the BIO_*printf functions. If applications use these functions in
this way then they could be vulnerable. OpenSSL itself uses these functions
when printing out human-readable dumps of ASN.1 data. Therefore applications
that print this data could be vulnerable if the data is from untrusted sources.
OpenSSL command line applications could also be vulnerable where they print out
ASN.1 data, or if untrusted data is passed as command line arguments. Libssl is
not considered directly vulnerable. Additionally certificates etc received via
remote connections via libssl are also unlikely to be able to trigger these
issues because of message size limits enforced within libssl.

CVE-2016-0797

In the BN_hex2bn function the number of hex digits is calculated using an int
value |i|. Later |bn_expand| is called with a value of |i * 4|. For large
values of |i| this can result in |bn_expand| not allocating any memory because
|i * 4| is negative. This can leave the internal BIGNUM data field as NULL
leading to a subsequent NULL ptr deref. For very large values of |i|, the
calculation |i * 4| could be a positive value smaller than |i|. In this case
memory is allocated to the internal BIGNUM data field, but it is insufficiently
sized leading to heap corruption. A similar issue exists in BN_dec2bn. This
could have security consequences if BN_hex2bn/BN_dec2bn is ever called by user
applications with very large untrusted hex/dec data. This is anticipated to be
a rare occurrence. All OpenSSL internal usage of these functions use data that
is not expected to be untrusted, e.g. config file data or application command
line arguments. If user developed applications generate config file data based
on untrusted data then it is possible that this could also lead to security
consequences. This is also anticipated to be rare.

CVE-2016-0798

The SRP user database lookup method SRP_VBASE_get_by_user had confusing memory
management semantics; the returned pointer was sometimes newly allocated, and
sometimes owned by the callee. The calling code has no way of distinguishing
these two cases. Specifically, SRP servers that configure a secret seed to hide
valid login information are vulnerable to a memory leak: an attacker connecting
with an invalid username can cause a memory leak of around 300 bytes per
connection. Servers that do not configure SRP, or configure SRP but do not
configure a seed are not vulnerable. In Apache, the seed directive is known as
SSLSRPUnknownUserSeed. To mitigate the memory leak, the seed handling in
SRP_VBASE_get_by_user is now disabled even if the user has configured a seed.
Applications are advised to migrate to SRP_VBASE_get1_by_user. However, note
that OpenSSL makes no strong guarantees about the indistinguishability of valid
and invalid logins. In particular, computations are currently not carried out
in constant time.

CVE-2016-0705

A double free bug was discovered when OpenSSL parses malformed DSA private keys
and could lead to a DoS attack or memory corruption for applications that
receive DSA private keys from untrusted sources. This scenario is considered
rare.

CVE-2016-0800

A cross-protocol attack was discovered that could lead to decryption of TLS
sessions by using a server supporting SSLv2 and EXPORT cipher suites as a
Bleichenbacher RSA padding oracle. Note that traffic between clients and non-
vulnerable servers can be decrypted provided another server supporting SSLv2
and EXPORT ciphers (even with a different protocol such as SMTP, IMAP or POP)
shares the RSA keys of the non-vulnerable server. This vulnerability is known
as DROWN (CVE-2016-0800). Recovering one session key requires the attacker to
perform approximately 2^50 computation, as well as thousands of connections to
the affected server. A more efficient variant of the DROWN attack exists
against unpatched OpenSSL servers using versions that predate 1.0.2a, 1.0.1m,
1.0.0r and 0.9.8zf released on 19/Mar/2015 (see CVE-2016-0703 below). Users can
avoid this issue by disabling the SSLv2 protocol in all their SSL/TLS servers,
if they've not done so already. Disabling all SSLv2 ciphers is also sufficient,
provided the patches for CVE-2015-3197 (fixed in OpenSSL 1.0.1r and 1.0.2f)
have been deployed. Servers that have not disabled the SSLv2 protocol, and are
not patched for CVE-2015-3197 are vulnerable to DROWN even if all SSLv2
ciphers are nominally disabled, because malicious clients can force the use of
SSLv2 with EXPORT ciphers. OpenSSL 1.0.2g and 1.0.1s deploy the following
mitigation against DROWN: SSLv2 is now by default disabled at build-time.
Builds that are not configured with "enable-ssl2" will not support SSLv2.
Even if "enable-ssl2" is used, users who want to negotiate SSLv2 via the
version-flexible SSLv23_method() will need to explicitly call either of:
SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2); or SSL_clear_options(ssl,
SSL_OP_NO_SSLv2); as appropriate. Even if either of those is used, or the
application explicitly uses the version-specific SSLv2_method() or its client
or server variants, SSLv2 ciphers vulnerable to exhaustive search key recovery
have been removed. Specifically, the SSLv2 40-bit EXPORT ciphers, and SSLv2
56-bit DES are no longer available. In addition, weak ciphers in SSLv3 and up
are now disabled in default builds of OpenSSL. Builds that are not configured
with "enable-weak-ssl-ciphers" will not provide any "EXPORT" or "LOW" strength
ciphers.

Signed-off-by: Jo-Philipp Wich <jow@openwrt.org>
Backport of r48868.

SVN-Revision: 48869

8 years agoCC: toolchain: use latest glibc 2.21 revision
John Crispin [Tue, 1 Mar 2016 08:53:27 +0000 (08:53 +0000)]
CC: toolchain: use latest glibc 2.21 revision
Fixes "CVE-2015-7547 --- glibc getaddrinfo() stack-based buffer
overflow"

Signed-off-by: Michael Marley <michael@michaelmarley.com>
SVN-Revision: 48864

8 years agokernel: add Realtek USB to Ethernet modules
John Crispin [Tue, 1 Mar 2016 08:15:04 +0000 (08:15 +0000)]
kernel: add Realtek USB to Ethernet modules

https://forum.openwrt.org/viewtopic.php?pid=279394

Signed-off-by: Dirk Neukirchen <dirkneukirchen@web.de>
Backport of r46031

SVN-Revision: 48863

8 years agobase-files: add /etc/profile.d support
John Crispin [Tue, 1 Mar 2016 08:15:04 +0000 (08:15 +0000)]
base-files: add /etc/profile.d support

OpenWrt should support an optional /etc/profile.d directory like
most other Linux distributions. This allows packages to install
their own scripts into /etc/profile.d/ directory.

The file suffix should make clear, that these scripts
are (sourced) shell-snippets. If the user needs e.g. php or lua,
one must make sure that the interpreter is called.
The reverse failsafe test makes sure, that the effective returncode is 0.

A typcal usecase is the inclusion of private helpers,
special variables or aliases, which at the moment needs
patching the sourcecode and is not well maintainable.
Now the builder can simply add there files.

v1 initial work of Hendrik Lüth <hendrik@linux-nerds.de>
v2 changes regarding RFC (e.g. thomas.langer@lantiq.com)
v3 changes regarding RFC (e.g. mschiffer@universe-factory.net)
v4 keep it simple and mimic OpenWrt style

Signed-off-by: Bastian Bittorf <bittorf@bluebottle.com>
Backport of r46965

SVN-Revision: 48862

8 years agox86/generic: enable pata_atiixp driver
John Crispin [Tue, 1 Mar 2016 08:15:03 +0000 (08:15 +0000)]
x86/generic: enable pata_atiixp driver

This enables booting from devices that use an ATI PATA controller for
the boot device, such as the embedded CF cards in Fujitsu-Siemens Futro
thin-clients.

Signed-off-by: Felix Kaechele <heffer@fedoraproject.org>
Backport of r47304

SVN-Revision: 48861

8 years agoar71xx: add support for TP-LINK TL-WR941ND v6 (international version)
John Crispin [Tue, 1 Mar 2016 08:15:02 +0000 (08:15 +0000)]
ar71xx: add support for TP-LINK TL-WR941ND v6 (international version)

The international version is completely different from the already
supported Chinese version. The WLAN of the QCA956x SoC used by this router
has been fixed in r46948.

This patch looks like it changes a lot in
700-MIPS-ath79-openwrt-machines.patch; that is not the case. Unfortunately,
quilt decided to completely reorganize the Kconfig patch even though only
a single section has been added.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Backport of r47420

SVN-Revision: 48860

8 years agoar71xx: rename firmware for Chinese version of TL-WR941ND v6
John Crispin [Tue, 1 Mar 2016 08:14:59 +0000 (08:14 +0000)]
ar71xx: rename firmware for Chinese version of TL-WR941ND v6

It seems there are two *different* versions on TL-WR941ND v6: Chinese
and European one. Currently supported by OpenWrt is Chinese one. Rename
firmware to make it clear.
For details see:
http://wiki.openwrt.org/toh/tp-link/tl-wr941nd
https://forum.openwrt.org/viewtopic.php?id=59295

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Backport of r47107

SVN-Revision: 48859

8 years agomac80211: set mcast-rate for mesh point interfaces
John Crispin [Tue, 1 Mar 2016 08:14:59 +0000 (08:14 +0000)]
mac80211: set mcast-rate for mesh point interfaces

In order to set the multicast rate for mesh point interfaces the "mesh join"
was made explicit and moved to mac80211_setup_vif(), similar to how it is
done for IBSS interfaces.

Previously, the mesh join was made implicit in case authentication (i.e.
$key) was not used when creating the interface in mac80211_prepare_vif(),
while using authentication would create the interface first, then join
later in mac80211_setup_vif() by starting authsae.

Signed-off-by: Nils Schneider <nils@nilsschneider.net>
Backport of r47408

SVN-Revision: 48858

8 years agoar71xx: compex wpj531 register usb support
John Crispin [Tue, 1 Mar 2016 08:14:58 +0000 (08:14 +0000)]
ar71xx: compex wpj531 register usb support

Signed-off-by: Christian Mehlis <christian@m3hlis.de>
Backport of r46283

SVN-Revision: 48857

8 years agoar71xx: compex wpj531 fix setting mac address on ethernet
John Crispin [Tue, 1 Mar 2016 08:14:57 +0000 (08:14 +0000)]
ar71xx: compex wpj531 fix setting mac address on ethernet

Signed-off-by: Christian Mehlis <christian@m3hlis.de>
Backport of r46282

SVN-Revision: 48856

8 years agoar71xx: compex wpj531 fix ethernet registration
John Crispin [Tue, 1 Mar 2016 08:14:56 +0000 (08:14 +0000)]
ar71xx: compex wpj531 fix ethernet registration

In fb6f62e97733312053ab593fcf68eea47a21169e several settings
are set on the ethernet device, but they are not working.

Fix Ethernet by setting the right values.

Signed-off-by: Christian Mehlis <christian@m3hlis.de>
Backport of r46281

SVN-Revision: 48855

8 years agoar71xx: rework patch for qca953x/956x
John Crispin [Tue, 1 Mar 2016 08:14:55 +0000 (08:14 +0000)]
ar71xx: rework patch for qca953x/956x

Patch cherry-picked from the following location:
https://www.codeaurora.org/cgit/quic/qsdk/oss/system/openwrt/commit/?h=release/coconut_ioe4531_2.0&id=5c357bf6c763e4140dddcc9a3bc5f005525a9c0e

Changelist,
    - add more register defines
    - add EHCI support
    - fix GPIO pin count to 18
    - fix chained irq disabled
    - fix GMAC0/GMAC1 initial
    - fix WMAC irq number to 47
    - merge the changes of dev-eth.c from the patch to file.

Signed-off-by: Miaoqing Pan <miaoqing@codeaurora.org>
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Backport of r46207

SVN-Revision: 48854

8 years agoar71xx: fix ethernet initialization on QCA953x-based boards (TP-Link TL-WR841N/ND...
John Crispin [Tue, 1 Mar 2016 08:14:53 +0000 (08:14 +0000)]
ar71xx: fix ethernet initialization on QCA953x-based boards (TP-Link TL-WR841N/ND v9, Compex WPJ531)

The initialization routines for these boards were relying on some (wrong)
defaults for the QCA953x ethernet. Make these defaults explicit to prevent
breaking them when the QCA953x defaults are fixed.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Backport of r46206

SVN-Revision: 48853

8 years agoar71xx/lzma-loader: fix O32 ABI conformance
John Crispin [Tue, 1 Mar 2016 08:14:52 +0000 (08:14 +0000)]
ar71xx/lzma-loader: fix O32 ABI conformance

According to the calling convention of the o32 ABI the caller
function must reserve stack space for $a0-$a3 registers in case
the callee needs to save its arguments.

The assembly code of the loader does not reserve stack space for
these registers thus when the 'loader_main' function needs to save
its arguments, those will be stored in the 'workspace' area instead
of the stack.

Because the workspace area is also used by other part of the code, the
saved register values gets overwritten and this often leads to failed
kernel boots.

Fix the code to reserve stack space for the registers to avoid this
error.

Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
Backport of r48279

SVN-Revision: 48852

8 years agokernel/gpio_keys: load module on pre-init
John Crispin [Tue, 1 Mar 2016 08:14:51 +0000 (08:14 +0000)]
kernel/gpio_keys: load module on pre-init

fix rescue mode on wdr4900

Signed-off-by: Alexander Couzens <lynxis@fe80.eu>
Backport of r48793

SVN-Revision: 48851

8 years agobrcm2708-gpu-fw: move pkg build dir to kernel build dir to fix rebuild bugs after...
John Crispin [Tue, 1 Mar 2016 08:14:50 +0000 (08:14 +0000)]
brcm2708-gpu-fw: move pkg build dir to kernel build dir to fix rebuild bugs after cleaning the kernel tree

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Backport of r48732

SVN-Revision: 48850

8 years agokernel: disable software protection bits for macronix flash chips at init
John Crispin [Tue, 1 Mar 2016 08:14:49 +0000 (08:14 +0000)]
kernel: disable software protection bits for macronix flash chips at init

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Backport r47626

SVN-Revision: 48849

8 years agofirmware-utils: mkfwimage: fix firmware_max_length for XM layout
John Crispin [Tue, 1 Mar 2016 08:14:48 +0000 (08:14 +0000)]
firmware-utils: mkfwimage: fix firmware_max_length for XM layout

The new u-boot version bundled with the 5.6.x firmwares from Ubiquiti gets
confused by the smaller rootfs partition size; this can lead to various
issues:

1. We've gotten reports that flashing from the 5.6.x stock firmware to
   OpenWrt will brick devices; I wasn't able to reproduce this myself
2. Flashing from 5.5.x stock firmware to OpenWrt and back to stock (via
   TFTP recovery), following by an update to 5.6.x via web interface can
   yield a bricked device with the following properties:
   - It can't be booted without entering commands over a serial console, as
     u-boot supplies the wrong MTD layout
   - The web interface won't accept any image with the original flash
     layout, so stock firmware upgrades are impossible
   - As the TFTP recovery doesn't update u-boot, returning to the old
     u-boot from firmware 5.5.x is impossible

To recover from 2., creating an OpenWrt image which doesn't set u-boot as
read-only and flashing a backup of the old u-boot from there is the only
way known to me. (Fixing the mtdparts variable in u-boot-env from OpenWrt
might also work; settings this from u-boot over serial didn't have
any permanent effect.)

Fix all of this by setting the correct flash layout also used by the stock
firmware. Flashing has been tested from both firmware 5.5.x and 5.6.x. The
fixed layout also matches the mtdparts defined by OpenWrt.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Backport of r48829

SVN-Revision: 48848

8 years agoar71xx, firmware-utils: split ubdev01 flash layout from XM
John Crispin [Tue, 1 Mar 2016 08:14:47 +0000 (08:14 +0000)]
ar71xx, firmware-utils: split ubdev01 flash layout from XM

The ubdev01 profile defines its own MTDPARTS with smaller firmware
partition, so give it its own UBNT_BOARD in mkfwimage.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Backport of r48828

SVN-Revision: 48847

8 years agofirmware-utils: mkfwimage: add -Wall, fix obvious bugs causing compile warnings
John Crispin [Tue, 1 Mar 2016 08:14:47 +0000 (08:14 +0000)]
firmware-utils: mkfwimage: add -Wall, fix obvious bugs causing compile warnings

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Backport of r48827

SVN-Revision: 48846

8 years agoOpenSSL: Added source/old to PKG_SOURCE_URL
John Crispin [Tue, 1 Mar 2016 08:14:46 +0000 (08:14 +0000)]
OpenSSL: Added source/old to PKG_SOURCE_URL

OpenSSL moves old versions of the library from
http://www.openssl.org/source/ to
http://www.openssl.org/source/old/$version/ breaking the old links.
That behavior breaks the OpenWRT-build every time OpenSSL releases
a new version.

This patch adds http://www.openssl.org/source/old/$version/ to the
PKG_SOURCE_URL of OpenSSL to avoid breaking the build whenever
OpenSSL releases a new version.

Signed-off-by: Kevin Kirsch <ranlvor@starletp9.de>
Reviewed-by: Alexander Dahl <post@lespocky.de>
Backport of r47860

SVN-Revision: 48845

8 years agoar71xx: fix AR71XX_MODEL on TP-Link TL-WR703N
John Crispin [Tue, 1 Mar 2016 08:14:45 +0000 (08:14 +0000)]
ar71xx: fix AR71XX_MODEL on TP-Link TL-WR703N

The hwid check was wrong, causing the AR71XX_MODEL value to end with a
space (as $hwver was unset).

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Backport of r47553

SVN-Revision: 48844

8 years agoar71xx: add support for the TP-LINK TL-WR1043ND v3
John Crispin [Tue, 1 Mar 2016 08:14:44 +0000 (08:14 +0000)]
ar71xx: add support for the TP-LINK TL-WR1043ND v3

The hardware of the v3 is identical to the v2.

Based-on-patch-by: Andreas Ziegler <github@andreas-ziegler.de>
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Backport of r47406

SVN-Revision: 48843

8 years agotools/firmware-utils: tplink-safeloader: add version 1.1 support to CPE210/220/510/520
John Crispin [Tue, 1 Mar 2016 08:14:43 +0000 (08:14 +0000)]
tools/firmware-utils: tplink-safeloader: add version 1.1 support to CPE210/220/510/520

The hardware is identical to version 1.0, add the new models to the support
list.

Also remove the empty line at the end of the support list, the current
stock images don't have it either.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Backport of r47212

SVN-Revision: 48842

8 years agotools/firmware-utils: tplink-safeloader: fix support-list format, clean up vendor...
John Crispin [Tue, 1 Mar 2016 08:14:43 +0000 (08:14 +0000)]
tools/firmware-utils: tplink-safeloader: fix support-list format, clean up vendor information

The first 4 bytes of the support list and the vendor information are
supposed to contain the length of these fields.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Backport of r47211

SVN-Revision: 48841

8 years agotools/firmware-utils: tplink-safeloader: clean up code
John Crispin [Tue, 1 Mar 2016 08:14:42 +0000 (08:14 +0000)]
tools/firmware-utils: tplink-safeloader: clean up code

There is no reason for the internal functions not to be static.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Backport of r47210

SVN-Revision: 48840

8 years agoodhcp6c: minor fixes
John Crispin [Tue, 1 Mar 2016 08:14:41 +0000 (08:14 +0000)]
odhcp6c: minor fixes

Better synchronize RA & DHCPv6 events
Accumulate some events to avoid flooding
Restart softwires for address and prefix changes

Signed-off-by: Steven Barth <steven@midlink.org>
Backport of r46518

SVN-Revision: 48839

8 years agobase-files: default_postinst: propagate the real postinst return code
John Crispin [Tue, 1 Mar 2016 08:14:40 +0000 (08:14 +0000)]
base-files: default_postinst: propagate the real postinst return code

Using the postinst script for sanity checks and expecting opkg to fail
if the postinst didn't return 0 was possible in Barrier Breaker, propagate
the real postinst return code through default_postinst to restore this
behaviour.

Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Backport of r46653

SVN-Revision: 48838