Add three download mirrors for 'bc', as the Makefile currently defines only
one download location.
@GNU can not be used, as the most recent version of 'bc' is not available at
the general GNU mirrors, and can only be found at the "gnu-alpha" mirrors.
Signed-off-by: Hannu Nyman <hannu.nyman at iki.fi>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 45908
VGV7519 is currently the only device with brnboot support. It seems to
be happy with 0x2083b8ed as CRC32 poly. However, VGV7510KW22 fails to
validate the checksum - it requires 0x04c11db7 instead.
I have built an brnboot image manually on the command line, once with
the old code and then with the new code but passing the old CRC32
poly value. Both resulted in a brnboot image with the same sha1sum.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
SVN-Revision: 45881
Simply disable the build of the "doc" and "tests" directories to speed up the
build and avoid the potential for errors coming from external tools being used
(e.g: source-highlight installed on a non-standard location).
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 45713
Fedora 22 includes gcc5 by default
build leads to error:
include/linux/compiler-gcc.h:114:30: fatal error: linux/compiler-gcc5.h: No such file or directory
backport upstream patch: "Add linux/compiler-gcc5.h to fix builds with gcc5"
id: 478b02f1a7043b673565075ea5016376f3293b23
Issue was reported at
https://forum.openwrt.org/viewtopic.php?id=56853
& replicated in a VM
created, flashed and tested ar71xx image successfully
as test
Signed-off-by: Dirk Neukirchen <dirkneukirchen@web.de>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 45703
The rework of the host tools discovery caused automake to embed
absolute paths to staging_dir/host/bin/perl into the shebang
of the generated automake executables.
Switch to the portable "/usr/bin/env perl" and enable global
warnings through "$^W" instead of the "-w" interpreter argument.
Signed-off-by: Jo-Philipp Wich <jow@openwrt.org>
SVN-Revision: 44972
The rework of the host tools discovery caused autoconf to embed
absolute paths to staging_dir/host/bin/perl into the shebang
of the generated autoconfig executables.
Switch to the portable "/usr/bin/env perl" and enable global
warnings through "$^W" instead of the "-w" interpreter argument.
Signed-off-by: Jo-Philipp Wich <jow@openwrt.org>
SVN-Revision: 44971
The feeds that use them have been unmaintained for a long time, and
something similar can easily be added as a regular host build package.
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
SVN-Revision: 44869
Bump GNU patch in tools from 2.7.1 to 2.7.5.
Change download URL to use @GNU alias.
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
SVN-Revision: 44760
OpenWrt hides verbose output by default,
regardless of automake silent-rules being en-/disabled.
If we enable verbose output for package builds (V=s)
however, we'd like to see as most as possible.
Signed-off-by: Mirko Vogt <mirko@openwrt.org>
SVN-Revision: 44685
Projects using silent-rules might otherwise fail to compile.
This is due to the following resulting Makefile code:
AM_V_P = $(am__v_P_$(V))
am__v_P_ = $(am__v_P_$(AM_DEFAULT_VERBOSITY))
am__v_P_0 = false
am__v_P_1 = :
Automake directly uses $(V) for variable name expansion
which fails if $(V) is set to sth. else other than '0' or '1'.
This patch forces automake to not take $(V) into account
but to always use the default config for verbosity.
A better approach would be to actually take $(V) into account,
however not just pass it through (AM_V='$(V)') but
set AM_V to 0 if $V=0 / unset and to 1 otherwise.
This plan however is foiled due to my frustration about
automake as well as my lack of m4 skills.
automake bug report: http://debbugs.gnu.org/cgi/bugreport.cgi?bug=20077
Signed-off-by: Mirko Vogt <mirko@openwrt.org>
SVN-Revision: 44684
We do not need to align the start of read only rootfs's to erase blocks.
This allows us to write the squashfs rootfs directly behind the kernel,
potentially freeing up one erase block.
We still need to align for jffs2, so add a flag for imagetag to
optionally align the rootfs start.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 44556
memset() was called with a size argument against a pointer size, not the
structure size itself.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 43913
The new TP-LINK Pharos series uses a new bootloader, the "TP-LINK Safeloader".
It uses an advanced firmware image format, containing an image partition table
and a flash partition table (and image partitions are mapped to the
corresponding flash partitions). The exact image format is documented in the
source code.
Furthermore, the bootloader expects the kernel image as an ELF executable.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
SVN-Revision: 43384
(Reposted due to an issue with the patchwork server during original submission)
Unbranded. Silkscreen on PCB is “A5-V11”, believed to be made by Bococom (or at least uses Bococom image encryption - as used on poray devices - but different key)
Signed-off-by: Gareth Bryan <gareth@mx9.org>
SVN-Revision: 43102
Compiling the host tools on the new x32 architecture (which is
an ILP32 ELF32 system on an amd64 CPU) fails for various reasons.
gmp: pull same fix I applied to OpenADK, which was inspired
by the fix in the Debian source package
mtd-utils: write a workaround myself; only affects x32, but
the use of llseek is dangerous according to the manpage, so
the guard ifdef should probably go away
findutils: pull fix straight from the Debian source packae
Signed-off-by: Thorsten Glaser <tg@mirbsd.org>
SVN-Revision: 43060
I defined a new download method @SAVANNAH in include/download.mk and scripts/download.pl,
and converted quilt and qemu to use that method.
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
SVN-Revision: 42840
I have never seen a LP-PHY with core rev 16 or higher, but the ucode
will be included, because we need LP-PHY 13 and 15 and N-PHY core rev
16. Comment out the code for now.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 41595
This adds support for new firmware files from b43 and selects the ucode
based on the PHY type now.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 41592
This patch adds factory image building for the DGN3500, all variants,
and fixes sysupgrade images to make them play nice with the sercomm
secondary boot loader.
The factory images can be used directly in the update dialog in the
interface of the stock firmware and via the special Sercomm bootmode
and a special windows flashing utility (allegedly present in the CD
that came with the device -- but it's also compatible with the NSLU2
Upgrade_207_XP utility.) The special bootmode can be activated by
turning the device on while holding the reset button pressed, then
releasing it when the power led starts blinking red and green. Please
notice that if using the 207 utility, it will always report that the
flashing failed even though it completed successfully. Just power
cycle the router manually after the utility reports the failure and
OpenWRT will boot. This same utility (despite reporting failure in
this case too) can revert a DGN3500 (any variant) to the appropriate
stock Netgear firmware.
This patch is a heavily modified version of a package I found on the
OpenWRT forum with a couple fixes and features added -- mainly the
generation of all the different image variants to support all known
models directly, atm known variants are AnnexA-WW, AnnexA-NA and
AnnexB-DE/GR.
I tested the patch successfully on my device.
Signed-off-by: Marco Antonio Mauro <marcus90@gmail.com>
SVN-Revision: 41236
This patch series is extracted from
http://ftp.de.debian.org/debian/pool/main/g/genext2fs/genext2fs_1.4.1-4.debian.tar.gz
The patches are used in Debian for quite a long time, so I assume that
this is solid material. At least, my Ubuntu host fsck.ext4 does not bark :-)
The goal is to allow building filesystems with larger blocksizes instead of the
current default of 1k. This should improve performance and lifetime when the
filesystem is stored e.g. on a SD card (on Raspberry Pi/I2SE Duckbill for example)
which uses internal flash memory. Writing to flash memory is slow because writing
the data of one block results in erasing a whole erase block of the flash memory.
Thus it is preferable to align the filesystem block size on a flash device with
the erase blocksize, or at least bring it closer to the later one, to avoid
unnecessary write amplification.
Signed-off-by: Michael Heimpold <mhei@heimpold.de>
SVN-Revision: 40921
