Commit Graph

51145 Commits

Author SHA1 Message Date
Rafał Miłecki
a6a0b252ba bcm4908: add sysupgrade support
It supports flashing OpenWrt images (bootfs & UBI upgrade) as well as
vendor images (whole MTD partition write).

Upgrading cferom is unsupported. It requires copying device specific
data (like MAC) to target image before flashing.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
2021-04-07 14:44:20 +02:00
Felix Fietkau
e48c6400e4 mt76: update to the latest version
7d35b7a15d1d mt76: mt7915: add wifi subsystem reset
04122c89749d mt76: fix rx amsdu subframe processing
5e764ec9bece mt76: mt7921: introduce MT_WFDMA_DUMMY_CR definition
cf0badbc0497 mt76: mt7921: fix inappropriate WoW setup with the missing ARP informaiton
f32a4e15f5b2 mt76: mt7921: fix the dwell time control
54f52771a04a mt76: mt7921: fix kernel crash when the firmware fails to download
97189d2a045b mt76: mt7921: fix the insmod hangs
dcdbd7c89cf5 mt76: mt7921: fix MT_PCIE_MAC_INT_ENABLE access
813db729c02f mt76: mt7921: reduce the data latency during hw scan

Signed-off-by: Felix Fietkau <nbd@nbd.name>
2021-04-07 10:59:11 +02:00
Rafał Miłecki
ca9b1f15c4 bcm4908: pad firmware image bootfs JFFS2 partition to 8 MiB
This way MTD "bootfs" partition will be always 8+ MiB. This should be
enough for any custom / future firmware to fit its bootfs (e.g. big
kernel) without having to repertition whole flash. That way we can
preserve UBI and its erase counters during sysupgrade.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
2021-04-07 09:36:47 +02:00
Rafał Miłecki
ed7edf88e2 firmware-utils: bcm4908img: extract bootfs without padding
JFFS2 bootfs partition in a BCM4908 image usually includes some padding.
For flashing it individually (writing to designed MTD partition) we want
just JFFS2 data.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
2021-04-07 08:02:41 +02:00
Rafał Miłecki
dcbde11af1 firmware-utils: bcm4908img: fix extracting cferom
Fix offset to extract proper data when image contains vendor header.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
2021-04-07 00:22:37 +02:00
Rafał Miłecki
e33957c241 firmware-utils: bcm4908img: support extracting bootfs & rootfs
It's required for upgrading firmware using single partitions instead of
just blindly writing whole image.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
2021-04-07 00:21:40 +02:00
Rafał Miłecki
5314cab729 firmware-utils: bcm4908img: replace size with offset
It's much easier to operate on BCM4908 image data with absolute offset
of each section stored. It doesn't require summing sizes over and over.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
2021-04-07 00:12:01 +02:00
Koen Vandeputte
f4f4259ca1 ath79: ar934x-nand: fix broken nand on kernel 5.10
Booting current 5.10.x testing kernel shows this:

[    0.335781] nand: device found, Manufacturer ID: 0xec, Chip ID: 0xf1
[    0.342280] nand: Samsung NAND 128MiB 3,3V 8-bit
[    0.346982] nand: 128 MiB, SLC, erase size: 128 KiB, page size: 2048,
OOB size: 64
[    0.354703] ar934x-nand 1b800200.nand: unknown ECC mode 2
[    0.360193] ar934x-nand 1b800200.nand: nand_scan failed, err:-22
[    0.366341] ar934x-nand: probe of 1b800200.nand failed with error -22

...

[    1.089264] /dev/root: Can't open blockdev
[    1.093482] VFS: Cannot open root device "(null)" or
unknown-block(0,0): error -6
[    1.101077] Please append a correct "root=" boot option; here are the
available partitions:
[    1.109555] 1f00              64 mtdblock0
[    1.109559]  (driver?)
[    1.116199] 1f01              48 mtdblock1
[    1.116202]  (driver?)
[    1.122841] 1f02               4 mtdblock2
[    1.122844]  (driver?)
[    1.129493] 1f03               4 mtdblock3
[    1.129497]  (driver?)
[    1.136138] 1f04               4 mtdblock4
[    1.136142]  (driver?)
[    1.142787] Kernel panic - not syncing: VFS: Unable to mount root fs
on unknown-block(0,0)
[    1.151168] Rebooting in 1 seconds..

The issue is caused by a wrong check for ECC engine type
on newer kernels.

Fix this.

Bootlog after:

[    0.335689] nand: device found, Manufacturer ID: 0xec, Chip ID: 0xf1
[    0.342192] nand: Samsung NAND 128MiB 3,3V 8-bit
[    0.346884] nand: 128 MiB, SLC, erase size: 128 KiB, page size: 2048,
OOB size: 64
[    0.354611] Scanning device for bad blocks
[    0.364924] random: fast init done
[    0.496641] 3 fixed-partitions partitions found on MTD device
ar934x-nand
[    0.503545] Creating 3 MTD partitions on "ar934x-nand":
[    0.508865] 0x000000000000-0x000000040000 : "booter"
[    0.515133] 0x000000040000-0x000000400000 : "kernel"
[    0.523265] 0x000000400000-0x000008000000 : "ubi"

Fixes: fb64e2c30f ("ath79: ar934x-nand: add kernel 5.10 compatibility")
Signed-off-by: Koen Vandeputte <koen.vandeputte@citymesh.com>
2021-04-06 17:00:07 +02:00
Rafał Miłecki
880c8b4422 bcm4908: rename bootfs dummy file to the 1-openwrt
The purpose of that dummy file is to make CFE work properly with OpenWrt
bootfs. CFE for some reason ignores JFFS2 files with ino 0.

Rename it to 1-openwrt so:
1. It's consistent with bcm63xx
2. It's OpenWrt specific so sysupgrade can distinguish it from vendor
   images

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
2021-04-06 00:07:12 +02:00
Hans Dedecker
d49f1acff1 glibc: update to latest git HEAD (BZ #27468, BZ #27511)
e78ea9bd26 Update Nios II libm-test-ulps.
98bb18f52a malloc: Fix a realloc crash with heap tagging [BZ 27468]
fc4ecce85b S390: Also check vector support in memmove ifunc-selector [BZ #27511]

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
2021-04-05 20:25:42 +02:00
Daniel Golle
5e7d6d5ba4
uboot-mediatek: update to v2021.04
Remove patches merged upstream and refresh the remaining ones.
Runtime tested on
 * Bananapi BPi-R64 (eMMC and SD Card)
 * Linksys E8450 (SPI-NAND)
 * Ubiquiti UniFi 6 LR (SPI-NOR)

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
2021-04-05 18:57:51 +01:00
Stijn Tintel
bcdf600fc5 lldpd: further size reductions
Size difference on mips_4kec:
Before: 120196
After:  120006

Closes https://github.com/openwrt/openwrt/pull/3823

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
Suggested-by: Lucian Cristian <lucian.cristian@gmail.com>
2021-04-05 18:50:17 +03:00
Stijn Tintel
e4d7e7b3e4 lldpd: enable LTO
Size difference on mips_4kec:
W/o LTO: 139674
W/ LTO:  120196

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
Suggested-by: Lucian Cristian <lucian.cristian@gmail.com>
2021-04-05 18:50:11 +03:00
Stijn Tintel
8946be0ab5 lldpd: bump to 1.0.9
Contains fixes related to CVE-2020-27827.

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
2021-04-05 18:27:42 +03:00
Stijn Tintel
c935c6ffb6 lldpd: add libcap dependency
Now that libcap is in OpenWrt base, we can drop our custom patch to
disable libcap support and have lldpd depend on it instead. This will
allow the monitor process to drop its privileges instead of running as
root, improving security.

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
2021-04-05 18:20:48 +03:00
Stijn Tintel
0f7f4de6ba libcap: bump to 2.48
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
Acked-by: Hauke Mehrtens <hauke@hauke-m.de>
2021-04-05 18:20:45 +03:00
Stijn Tintel
dd91ba0d62 libcap: drop invalid copyright header
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
Acked-by: Hauke Mehrtens <hauke@hauke-m.de>
2021-04-05 18:20:40 +03:00
Stijn Tintel
427acb71fc libcap: import from packages feed
Having libcap in OpenWrt base allows us to enable libcap support in
other packages in base.

In lldpd, this would allow the monitor process to drop its privileges
instead of running as root, improving security. It will also allow us to
drop our patch to disable libcap.

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
Acked-by: Hauke Mehrtens <hauke@hauke-m.de>
2021-04-05 18:20:29 +03:00
Rafał Miłecki
180c463526 bcm4908: enable JFFS2 support
It's needed for accessing JFFS2 bootfs partition.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
2021-04-05 10:44:46 +02:00
Rafał Miłecki
ed847ef5f3 firmware-utils: bcm4908img: add bootfs support
This adds support for accessing bootfs JFFS2 partition in the BCM4908
image. Support includes:
1. Listing files
2. Renaming file (requires unchanged name length)

Above commands are useful for flashing BCM4908 images which by defualt
come with cferom.000 file and require renaming it.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
2021-04-05 10:13:05 +02:00
Rafał Miłecki
9c039d56a1 firmware-utils: bcm4908img: support extracting image data
It's useful for upgrading cferom, firmware, etc.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
2021-04-05 10:05:25 +02:00
Rafał Miłecki
6af45b842b firmware-utils: bcm4908img: find cferom size
It's important for modifying / extracting firmware content. cferom is
optional image content at the file beginning.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
2021-04-05 10:05:09 +02:00
Rafał Miłecki
9b9184f178 firmware-utils: bcm4908img: use "info" command displaying file info
BCM4908 image format contains some info that may be useful for info /
debugging purposes.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
2021-04-05 10:03:59 +02:00
Rafał Miłecki
d533b27bc0 firmware-utils: bcm4908img: support reading from stdin
1. Don't allow pipe stdin as we need to fseek()
2. Don't alow TTY as it doesn't make sense for binary input

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
2021-04-05 09:54:10 +02:00
Rafał Miłecki
a39f85d8b6 firmware-utils: bcm4908img: detect Netgear vendor firmware
Netgear uses CHK header which needs to be skipped when validating
BCM4908 image. Detect it directly in the bcm4908img tool. Dealing with
binary structs and endianess is way simpler in C.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
2021-04-05 09:54:10 +02:00
Rafał Miłecki
7d5f743942 firmware-utils: bcm4908img: extract parsing code
Move code parsing existing firmware file to separated function. This
cleans up existing code and allows reusing parsing code for other
commands.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
2021-04-05 09:54:10 +02:00
Robert Marko
faea7becaf
ipq40xx: add MikroTik hAP ac2 support
This adds support for the MikroTik RouterBOARD RBD52G-5HacD2HnD-TC
(hAP ac²), a  indoor dual band, dual-radio 802.11ac
wireless AP with integrated omnidirectional antennae, USB port and  five
10/100/1000 Mbps Ethernet ports.

See https://mikrotik.com/product/hap_ac2 for more info.

Specifications:
 - SoC: Qualcomm Atheros IPQ4018
 - RAM: 128 MB
 - Storage: 16 MB NOR
 - Wireless:
   · Built-in IPQ4018 (SoC) 802.11b/g/n 2x2:2, 2.5 dBi antennae
   · Built-in IPQ4018 (SoC) 802.11a/n/ac 2x2:2, 2.5 dBi antennae
 - Ethernet: Built-in IPQ4018 (SoC, QCA8075) , 5x 1000/100/10 port,
             passive PoE in
- 1x USB Type A port

Installation:
Boot the initramfs image via TFTP and then flash the sysupgrade
image using "sysupgrade -n"

Signed-off-by: Robert Marko <robimarko@gmail.com>
2021-04-05 04:13:28 +02:00
Daniel Golle
1db4681617 uboot-mediatek: bpi-r64: no longer force recovery on first-boot
Now that sdcard.gz image contains everything needed to boot straight
into production image, no longer force booting into recovery image on
first boot by removing the logic which implemented that.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
2021-04-04 10:51:25 +01:00
Felix Fietkau
0b0c94d152 mt76: update to the latest version
465dda65ee84 mt7615,mt7915: replace fw log 0-terminating code with wiphy info length limit
62b13f5352b8 mt76: mt7921: fix key set/delete issue
0ff3a336a8d8 mt7615,mt7915: fix a compiler warning
113ba8a81d54 mt76: mt7615: remove redundant dev_err call in mt7622_wmac_probe()
be1ab3b9ae7c mt76: mt7921: fix typo in mt7921_pci_resume
4e22f0dc934b mt76: mt7915: fix txpower init for TSSI off chips
e66a0b9b8d66 mt76: mt7615: always wake the device in mt7615_remove_interface
38f656768a90 mt76: mt7921: always wake the device in mt7921_remove_interface
6ee4770de083 mt76: mt7921: rework mt7921_mcu_debug_msg_event routine
e578b4b8d56a mt76: mt7615: fix .add_beacon_offload()
f8c6c7cbf10f mt76: mt7915: fix mt7915_mcu_add_beacon

Signed-off-by: Felix Fietkau <nbd@nbd.name>
2021-04-04 11:29:36 +02:00
Kevin Darbyshire-Bryant
438e88e672 kernel: bump 5.10 to 5.10.27
Use update_kernel to refresh all patches, required manual updates to:

610-netfilter_match_bypass_default_checks.patch
611-netfilter_match_bypass_default_table.patch
762-net-bridge-switchdev-Refactor-br_switchdev_fdb_notif.patch
764-net-bridge-switchdev-Send-FDB-notifications-for-host.patch

Run-tested: x86_64

Nothing screamed out but any funny business with linux bridging should
suspect this update first.

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
2021-04-03 21:07:22 +01:00
Hauke Mehrtens
b974293efa netifd: update to Git version 2021-04-03
f8899b9 netifd: bridge: set default value for igmp_snoop
327da98 netifd: add possibility to switch off route config

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
2021-04-03 19:11:38 +02:00
Hauke Mehrtens
1371910b76 uclient: update to Git version 2021-04-03
83efca2 tests: fix possibly longer start of HTTP server
64e00d6 uclient-fetch: document missing options

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
2021-04-03 19:06:09 +02:00
Rui Salvaterra
d38f456582 hostapd: enable airtime policy for the -basic variants
Airtime policy configuration is extremely useful in multiple BSS scenarios.
Since nowadays most people configure both private and guest networks (at
least), it makes sense to enable it by default, except for the most limited
of the variants.

Size of the hostapd-basic-openssl binary (mipsel 24Kc -O2):
543944 bytes (airtime policy disabled)
548040 bytes (airtime policy enabled)

Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
Acked-by: Daniel Golle <daniel@makrotopia.org>
2021-04-03 18:57:13 +02:00
Stijn Segers
1601b39b61 realtek: rename partitions in Netgear DTSI
Switch the Netgear DTSI for the Realtek target from the OEM partition
naming scheme to accepted OpenWrt naming practices. A quick git grep for
'u-boot-env' e.g. in the OpenWrt tree turns up almost 500 hits whereas
grepping for 'bdinfo' (the OEM equivalent) returns a meagre 14.

Signed-off-by: Stijn Segers <foss@volatilesystems.org>
2021-04-03 18:57:13 +02:00
Rui Salvaterra
565dfeb128 zram-swap: bail out early if the kernel doesn't support swap
Since KERNEL_SWAP is only enabled by default for !SMALL_FLASH targets, we need
to check if the current kernel supports swap before trying to configure
zram-swap, as opkg can't check for kernel dependencies.

Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
2021-04-03 18:57:13 +02:00
Rui Salvaterra
829fa33899 zram-swap: clean up the makefile
Break dependencies into separate lines, to improve the readability. Trim
trailing whitespace.

Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
2021-04-03 18:57:13 +02:00
Andreas Böhler
73ec21cde2 lantiq: add basic support for AVM FRITZ!Box 3390
The FRITZ!Box 3390 actually contains two SoCs, one Lantiq with a
5GHz WiFi and one AR9342 with a 2.4GHz WiFi. Only the Lantiq
has access to the flash memory, the Atheros runs fully from RAM.

Specifications
--------------

  - Lantiq 500 MHz
  - 128MiB RAM
  - 128MiB NAND
  - 256k Flash
  - AR9580 5GHz WiFi
  - AR9342 560 MHz
  - 64MiB RAM
  - AR9328 2.4GHz WiFi

Remarks
-------

This commit only adds support for the Lantiq side of things and
prepares the drivers for communication with the Atheros SoC. Thus,
only 5GHz WiFi works by default, the 2.4GHz WiFi will be added via
another target.

Some kernel patches will be required to add support for the Atheros SoC.

Installation
------------

Use the eva_ramboot.py script to boot the initramfs image. Then, transfer
the sysupgrade image to the device and run sysupgrade to flash it to the
NAND.

Signed-off-by: Andreas Böhler <dev@aboehler.at>
Acked-by: Aleksander Jan Bajkowski A.Bajkowski@stud.elka.pw.edu.pl
Signed-off-by: Joachim Cerny <cocktail_yogi@web.de>
2021-04-03 18:56:51 +02:00
Joachim Cerny
754eb75a8c lantiq: VR9: fix gpio-hog by defining the GPIO ranges
defined gpio-ranges according to Reference: https://github.com/torvalds/linux/blob/master/drivers/pinctrl/pinctrl-xway.c#L864
This change is needed by gpio-hog on Fritzbox 3390
thx to @abajk

Acked-by: Aleksander Jan Bajkowski A.Bajkowski@stud.elka.pw.edu.pl
Signed-off-by: Andreas Böhler <dev@aboehler.at>
Signed-off-by: Joachim Cerny <cocktail_yogi@web.de>
2021-04-03 18:56:02 +02:00
Jeroen Peelaerts
4f27ea7c33 lantiq: use ActualNetDataRate for speed reporting
Switch to Actual Net Data Rate (ACTNDR) for speed reporting on lantiq VDSL modems

Refer to ITU-T G.997.1 chapter 7.5.2.8

Independent whether retransmission is used or not in a given transmit direction:
-   In L0 state, this parameter reports the Net Data Rate (as specified in G.992.3, G.992.5 or G.993.2) at which the bearer channel is operating.
-   In  L2 state, the parameter contains the Net Data Rate (as specified in G.992.3, G.992.5 or G.993.2) in the previous L0 state.

Signed-off-by: Jeroen Peelaerts <jeroen.peelaerts@gmail.com>
Reviewed-by: Andre Heider <a.heider@gmail.com>
2021-04-03 18:56:02 +02:00
Jeroen Peelaerts
48162e4c0c lantiq: enable G.INP retransmission counters
This commit adds monitoring for a couple of DSL line features that are
present in the lantiq firmware blobs.

* G.INP ON/OFF
* Trellis encoding ON/OFF
* Virtaul Noise ON/OFF
* Bitswap ON/OFF

Difference in size for ltq-vdsl-app = 1k
Difference in size for kmod-ltq-vdsl-vr9 < 1k

Reviewed-by: Andre Heider <a.heider@gmail.com>
Signed-off-by: Jeroen Peelaerts <jeroen.peelaerts@gmail.com>
2021-04-03 18:56:02 +02:00
Shiji Yang
dfc9044fad ramips: fix IMAGE_SIZE of HC5x6: fix image size of HC5x61
"firmware" partition size defined in the device tree file is 0xf70000,
so the right IMAGE_SIZE is 15808k

Fixes: df1e5d6463 ("ramips: fix partition layout of hiwifi hc5x61")
Signed-off-by: Shiji Yang <yangshiji66@qq.com>
2021-04-03 18:56:02 +02:00
Daniel Engberg
e83f7e5d76 ramips: Fix booting on MQmaker WiTi board
This fixes the dreaded "lzma error 1" also reported on similar devices
Ref: https://bugs.openwrt.org/index.php?do=details&task_id=3057

Fixes: FS#3057
Signed-off-by: Daniel Engberg <daniel.engberg.lists@pyret.net>
2021-04-03 18:56:02 +02:00
René van Dorst
12f3d1466a ramips: Fix booting on MTC WR1201
This fixes the dreaded "lzma error 1" also reported on similar devices
Ref: https://bugs.openwrt.org/index.php?do=details&task_id=3057

Fixes: FS#3057
Signed-off-by: René van Dorst <opensource@vdorst.com>
2021-04-03 18:56:02 +02:00
Aleksander Jan Bajkowski
f777884ac4 armvirt: add 5.10 testing kernel
Add 5.10 as testing kernel. Builds successfully with
CONFIG_ALL_KMODS=y enabled.

Signed-off-by: Aleksander Jan Bajkowski <A.Bajkowski@stud.elka.pw.edu.pl>
2021-04-03 18:56:02 +02:00
Aleksander Jan Bajkowski
d9fa0daef1 armvirt: refresh 5.10 config
This was done by executing these commands:
$ make kernel_oldconfig CONFIG_TARGET=subtarget
$ make kernel_oldconfig CONFIG_TARGET=subtarget_platform

Some common symbols have been moved to target config.

Signed-off-by: Aleksander Jan Bajkowski <A.Bajkowski@stud.elka.pw.edu.pl>
2021-04-03 18:56:02 +02:00
Aleksander Jan Bajkowski
072af549fd armvirt: copy config files for 5.10
Copy config and patches for 5.10.

Signed-off-by: Aleksander Jan Bajkowski <A.Bajkowski@stud.elka.pw.edu.pl>
2021-04-03 18:56:02 +02:00
Aleksander Jan Bajkowski
1cadd8fb78 armvirt: refresh 5.4 config
This was done by executing these commands:
$ make kernel_oldconfig CONFIG_TARGET=subtarget
$ make kernel_oldconfig CONFIG_TARGET=subtarget_platform

Signed-off-by: Aleksander Jan Bajkowski <A.Bajkowski@stud.elka.pw.edu.pl>
2021-04-03 18:56:02 +02:00
Alexander Egorenkov
9318f61556 base-files: fix status display command
If service() is called w/o parameter then the status display for services
with multiple instances is incorrect. E.g. samba4 or wpad have 2 instances.

root@OpenWrt:~# /etc/init.d/samba4 status
running
root@OpenWrt:~# /etc/init.d/wpad status
running

Before change:
/etc/init.d/samba4                 enabled         stopped
/etc/init.d/wpad                   enabled         stopped

After change:
/etc/init.d/samba4                 enabled         running
/etc/init.d/wpad                   enabled         running

Signed-off-by: Alexander Egorenkov <egorenar-dev@posteo.net>
2021-04-03 18:56:02 +02:00
Tony Ambardar
aab3a04ce8 iproute2: fix libbpf detection with NLS enabled
Upstream iproute2 detects libbpf using a one-line $CC test-compile, which
normally ignores LDFLAGS. With NLS enabled however, LDFLAGS includes an
"rpath-link" linker option needed to resolve libintl.so. Its absence
causes both the compile and libbpf detection to fail:

  ld: warning: libintl.so.8, needed by libbpf.so, not found (try using
      -rpath or -rpath-link)
  ld: libelf.so.1: undefined reference to `libintl_dgettext'
  collect2: error: ld returned 1 exit status

Fix this by directly including $LDFLAGS in the test-compile command.

Reported-by: Ian Cooper <iancooper@hotmail.com>
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
2021-04-03 18:56:02 +02:00
Tony Ambardar
9a59f62f61 binutils: fix libbfd missing DSO dependency if NLS enabled
The libbfd package definition uses $(ICONV_DEPENDS) and $(INTL_DEPENDS)
but links against neither, leading to libbfd detection failures in other
packages (e.g. bpftools) and on-target relocation problems with libintl.so:

  root@OpenWrt:/# ldd /usr/lib/libbfd.so
        ldd (0x77db6000)
        libc.so => ldd (0x77db6000)
        libgcc_s.so.1 => /lib/libgcc_s.so.1 (0x77c6d000)
  Error relocating /usr/lib/libbfd.so: libintl_dgettext: symbol not found

Add NLS-conditional linking of "libintl" to fix this. Also remove libbfd
package dependency $(ICONV_DEPENDS) which is not used during building or
linking.

Tested with QEMU on malta/be32, after building all packages from binutils,
bpftools and iproute2, using different libc options musl and glibc.

Fixes: 08e8175696 ("binutils: use nls.mk to fix libbfd link errors in
other packages")
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
2021-04-03 18:56:02 +02:00