Only include the major version in the base image name or aliases.

The D-Link DIR-825 has only 6M of usable firmware space.

target_config.lua and target_config_check.lua don't pass a table of
callbacks anymore, so target_config_lib.lua can by simplified by moving
all the code that was in the returned function to the toplevel.

So far, we were using a sort operation on the generated .config to
implement precedence of =y packages over =m, and =m over unset.
Unfortunately, this sort not only used for packages, but for all config
lines. This made it impossible to override settings from targets/generic
in a target config when the new setting was sorted before the generic
setting.

To fix this, track configurations by their keys, so we can properly
override config keys that were set before. Value-based precedence is
only preserved for package configuration.

The config() and try_config() calls always take key and value as
separate arguments now. Strings are quoted automatically; the values
true, nil and false map to y, m and unset for tristate options. config()
can take an optional third argument to override the error message to
display when the setting fails to apply.

All existing target configs generate the same .config with the old and the
new code. The new code is also a bit faster on targets with many devices.

This commit assigns class-flags to devices. The following scheme is
used:

 - ath9k & ath10k: tiny if RAM <128M
 - ath10k & ath10k: tiny if RAM <256M
 - all: tiny if RAM <64M
 - all: tiny if flash <8M

All other devices automatically have the default device-class selected.

Remove the ath10k-ct-smallbuffers variant from the default
package-seleciton as it clashes with the QCA ath10k driver.

a regular Netgear WNDR3800
but identifies itself as wndr3800chmychart.

As upstream now has model detection for the Ubiquiti UniFi AC LR, we
need to provide an autoupdater image matching the expected filename.

Closes #1834

Their kernel partition is too small, the OpenWrt build system
is unable to create an image.

This changes the ath10k firmware and driver selection, as
ath10k-ct is now the default in upstream OpenWrt.

However, for 802.11s operation we need the QCA firmware and driver.

This adds previously absent ath10k packages definitions. This way,
devices correctly select CandelaTech firmware correctly, when
GLUON_WLAN_MESH is set to IBSS.

This is also necessary for all devices, when OpenWrt is switched to
19.07.

fixes #1834
fixes #1332

The OCEDO Koala was missing the correct package definition. Because of
this, firmware is potentially built with the wrong ath10k firmware /
driver.

The kernel partition on these devices is too small for the
OpenWrt 19.07 kernel. It is also located after the rootfs, so
we cannot grow it further.

Therefore mark these devices as deprecated now.

This reverts commit fb08e104.

Devices featuring ath10k radios should at least 128 MB of memory
as their driver is quite memory hungry.

Devices with a configuration (1x ath10k, 64M memory) like this have
often been seen as running out of memory.

The profile for GL.iNet AR750 currently selects ATH10K_PACKAGES, but this device actually has a QCA9887 which needs another driver.
This commit fixes the issue by setting ATH10K_PACKAGES_QCA9887 instead.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>

memory usage seems to behave more reasonable for a while now

The old bash-based parsing code was way too complex. Replace it with Lua.

This device doesn't have an Ethernet port, and even with the Ethernet
shield, eth0 is not configured.

For some reason, some of the target files were executable.

also added manifest aliases for gl-ar150, gl-ar300m, gl-ar750

this matches the output of
lua -e 'print(require(platform_info).get_image_name())'

This way the manifest file is accepted by the autoupdater

Signed-off-by: Martin Weinelt <martin@darmstadt.freifunk.net>

Signed-off-by: Christoph Krapp <achterin@googlemail.com>

Signed-off-by: Christoph Krapp <achterin@googlemail.com>

Fixes #856

This adds support for the TP-Link TL-WR902Ac v1 travel router.

The device is marked as broken due to 64MB which might be insufficient
in certain environments.

Some 32 MB devices like the Nanostation M2 suffer from sudden high loads
combined with a squashfs related OOM reboot:

  logd invoked oom-killer: gfp_mask=0x2420848, order=0, oom_score_adj=0
  CPU: 0 PID: 774 Comm: logd Not tainted 4.4.135 #0
  Stack : 804214dc 00000000 00000001 80480000 8182fa3c 80474803 804028d0 00000306
          804e378c 00001ade 00000040 00000000 00000000 800a7f10 00000006 00000000
          00000000 00000000 804063e0 80c69994 804e6542 800a5e8c 02420848 00000000
          00000001 801fd600 00000000 00000000 00000000 00000000 00000000 00000000
          00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
          ...
  Call Trace:
  [<800721cc>] show_stack+0x54/0x88
  [<800d5468>] dump_header.isra.4+0x48/0x130
  [<800d5c38>] check_panic_on_oom+0x48/0x84
  [<800d5d64>] out_of_memory+0xf0/0x324
  [<800d9888>] __alloc_pages_nodemask+0x6b8/0x724
  [<800d2960>] pagecache_get_page+0x154/0x270
  [<80134cb0>] __getblk_slow+0x15c/0x374
  [<80160418>] squashfs_read_data+0x1c8/0x6e8
  [<80164628>] squashfs_readpage_block+0x32c/0x4d8
  [<801622a4>] squashfs_readpage+0x5bc/0x6d0
  [<800dd030>] __do_page_cache_readahead+0x1f8/0x264
  [<800d479c>] filemap_fault+0x1a8/0x458
  [<800efc1c>] __do_fault+0x64/0xd0
  [<800f2824>] handle_mm_fault+0x4a4/0xb40
  [<80076e98>] __do_page_fault+0x134/0x470
  [<80060820>] ret_from_exception+0x0/0x10

Reduction of the SquashFS blocksize should reduce the problem slightly on
these nodes. The image size will increase slightly but this should be no
problem for ar71xx-generic (in contrast to ar71xx-tiny).

removes cryptic debug messages from kernel ring buffer and
also reduces memory usage by as much as 20-25 MiB.