[PATCH] ramips: fix Unifi 6 Lite boot failure with v5.15 kernels
Robert Marko
robimarko at gmail.com
Thu Nov 3 11:17:58 PDT 2022
On Thu, 3 Nov 2022 at 19:11, Christian Marangi <ansuelsmth at gmail.com> wrote:
>
> On Thu, Nov 03, 2022 at 06:54:47PM +0100, Bjørn Mork wrote:
> > The Unifi 6 Lite U-Boot does not relocate the Device Tree Blobs
> > found in the FIT image. It behaves as if fdt_high is set to
> > 0xffffffff whether or not this variable is defined.
> >
> > Kernel commit 79edff12060f ("scripts/dtc: Update to upstream
> > version v1.6.0-51-g183df9e9c2b9"), included in v5.12, imported
> > dtc commit 5e735860c478 ("libfdt: Check for 8-byte address
> > alignment in fdt_ro_probe_()") into the kernel
> >
> > The result is that we must ensure that the "data" property of
> > all the "flat_dt" type images is 8 byte aligned inside the FIT
> >
> > There is currently no provision in mkimage or dtc for aligning
> > properties with the blob. They are naturally 4 byte aligned.
> > Giving us 50% probability of a soft bricking image.
> >
> > Work around the missing aligment features in mkimage and dtc
> > by padding the kernel image and rerunning mkimage if the first
> > attempt failed.
> >
> > Example boot log when using a misaligned FIT image:
> >
> > reading kernel 0 from: 0x1d0000, size: 0x002d5000
> > Using 'config at 1' configuration
> > Verifying Hash Integrity ... OK
> > Trying 'kernel-1' kernel subimage
> > Description: MIPS OpenWrt Linux-5.15.76
> > Type: Kernel Image
> > Compression: lzma compressed
> > Data Start: 0x860000e4
> > Data Size: 2956005 Bytes = 2.8 MiB
> > Architecture: MIPS
> > OS: Linux
> > Load Address: 0x80001000
> > Entry Point: 0x80001000
> > Hash algo: crc32
> > Hash value: e1bc9460
> > Hash algo: sha1
> > Hash value: 6510c4ada31aeea81f2e8e537f78cb367e1c7fab
> > Verifying Hash Integrity ... crc32+ sha1+ OK
> > Using 'config at 1' configuration
> > Trying 'fdt-1' fdt subimage
> > Description: MIPS OpenWrt ubnt_unifi-6-lite device tree blob
> > Type: Flat Device Tree
> > Compression: uncompressed
> > Data Start: 0x862d1d0c
> > Data Size: 11387 Bytes = 11.1 KiB
> > Architecture: MIPS
> > Hash algo: crc32
> > Hash value: 16bb5a14
> > Hash algo: sha1
> > Hash value: 40f26bf28e33bbe661fec716929f2003301f5e4d
> > Verifying Hash Integrity ... crc32+ sha1+ OK
> > Booting using the fdt blob at 0x862d1d0c
> > Uncompressing Kernel Image ... OK
> > Using Device Tree in place at 862d1d0c, end 862d7986
> > [ 0.000000] Linux version 5.15.76 (bjorn at canardo) (mipsel-openwrt-linux-musl-gcc (OpenWrt GCC 11.3.0 r21167-1673b7dca384) 11.3.0, GNU ld (GNU Binutils) 2.37) #0 SMP Wed Nov 2 15:53:34 2022
> > [ 0.000000] SoC Type: MediaTek MT7621 ver:1 eco:3
> > [ 0.000000] printk: bootconsole [early0] enabled
> > [ 0.000000] CPU0 revision is: 0001992f (MIPS 1004Kc)
> > [ 0.000000] Initrd not found or empty - disabling initrd
> > [ 0.000000] VPE topology {2,2} total 4
> > [ 0.000000] Primary instruction cache 32kB, VIPT, 4-way, linesize 32 bytes.
> > [ 0.000000] Primary data cache 32kB, 4-way, PIPT, no aliases, linesize 32 bytes
> > [ 0.000000] MIPS secondary cache 256kB, 8-way, linesize 32 bytes.
> > [ 0.000000] Zone ranges:
> > [ 0.000000] Normal [mem 0x0000000000000000-0x000000000fffffff]
> > [ 0.000000] HighMem empty
> > [ 0.000000] Movable zone start for each node
> > [ 0.000000] Early memory node ranges
> > [ 0.000000] node 0: [mem 0x0000000000000000-0x000000000fffffff]
> > [ 0.000000] Initmem setup node 0 [mem 0x0000000000000000-0x000000000fffffff]
> > [ 0.000000] OF: fdt: No valid device tree found, continuing without
> > [ 0.000000] percpu: Embedded 11 pages/cpu s15632 r8192 d21232 u45056
> > [ 0.000000] Built 1 zonelists, mobility grouping on. Total pages: 64960
> > [ 0.000000] Kernel command line: rootfstype=squashfs,jffs2
> > [ 0.000000] Dentry cache hash table entries: 32768 (order: 5, 131072 bytes, linear)
> > [ 0.000000] Inode-cache hash table entries: 16384 (order: 4, 65536 bytes, linear)
> > [ 0.000000] Writing ErrCtl register=0004a000
> > [ 0.000000] Readback ErrCtl register=0004a000
> > [ 0.000000] mem auto-init: stack:off, heap alloc:off, heap free:off
> > [ 0.000000] Memory: 248504K/262144K available (7250K kernel code, 660K rwdata, 1536K rodata, 1228K init, 242K bss, 13640K reserved, 0K cma-reserved, 0K highmem)
> > [ 0.000000] SLUB: HWalign=32, Order=0-3, MinObjects=0, CPUs=4, Nodes=1
> > [ 0.000000] rcu: Hierarchical RCU implementation.
> > [ 0.000000] Tracing variant of Tasks RCU enabled.
> > [ 0.000000] rcu: RCU calculated value of scheduler-enlistment delay is 10 jiffies.
> > [ 0.000000] NR_IRQS: 256
> > [ 0.000000] Kernel panic - not syncing: Failed to find mediatek,mt7621-sysc node
> > [ 0.000000] Rebooting in 1 seconds..
> > [ 0.000000] Reboot failed -- System halted
> >
> > Signed-off-by: Bjørn Mork <bjorn at mork.no>
> > ---
> > The code below will probably look stupid to anyone with a clue. but I
> > hope it serves as an illustration of what we need to do. I'd appreciate
> > it if anyone can teach me
> > - how to do alignment inside a FIT using mkimage, or
> > - how to find the position of the embedded fdt blobs in the FIT, or
> > - how to check whether a value is aligned
> >
> > In any case, believe fixing this ASAP is critical. As it is now, any
> > image with a 5.15 kernel is just as likely to brick the device as is to
> > boot.
>
> If it's critical enough should we revert the affected commit just in
> case while we understand this problem?
> Consider that sooner or later we will have to fix it as 6.1 will have
> this problem. But if this is present on also other target we should
> operate on a more generic way than apply a workaround specific to a
> device.
This is a broken vendor bootloader doing its thing, so we shouldn't
revert upstream commits
that is just enforcing the boot specification.
This was introduced in 5.12 anyway, so most of the broken devices
should have already surfaced.
What should be done is to expand mkimage with an option to align to an
8-byte boundary.
Regards,
Robert
>
> >
> >
> > target/linux/ramips/image/mt7621.mk | 19 ++++++++++++++++++-
> > 1 file changed, 18 insertions(+), 1 deletion(-)
> >
> > diff --git a/target/linux/ramips/image/mt7621.mk b/target/linux/ramips/image/mt7621.mk
> > index 3ef4cf4efb8f..e3b11abd14ce 100644
> > --- a/target/linux/ramips/image/mt7621.mk
> > +++ b/target/linux/ramips/image/mt7621.mk
> > @@ -132,6 +132,23 @@ define Build/zyxel-nwa-fit
> > @mv $@.new $@
> > endef
> >
> > +define Build/fit-dtalign
> > + $(TOPDIR)/scripts/mkits.sh \
> > + -D $(DEVICE_NAME) -o $@.its -k $@ \
> > + -C $(word 1,$(1)) \
> > + -d $(word 2,$(1)) \
> > + -a $(KERNEL_LOADADDR) -e $(if $(KERNEL_ENTRY),$(KERNEL_ENTRY),$(KERNEL_LOADADDR)) \
> > + -c $(if $(DEVICE_DTS_CONFIG),$(DEVICE_DTS_CONFIG),"config-1") \
> > + -A $(LINUX_KARCH) -v $(LINUX_VERSION)
> > + PATH=$(LINUX_DIR)/scripts/dtc:$(PATH) mkimage -f $@.its $@.new
> > + ( pos=$$(grep -aob $$(echo -ne "\xd0\x0d\xfe\xed") $@.new | cut -f1 -d: | tail -1); \
> > + if [ "$$pos" -ne "$$((8*($$pos/8)))" ]; then \
> > + dd if=/dev/zero bs=4 count=1 >> $@; \
> > + PATH=$(LINUX_DIR)/scripts/dtc:$(PATH) mkimage -f $@.its $@.new; \
> > + fi )
> > + @mv $@.new $@
> > +endef
> > +
> > define Device/dsa-migration
> > DEVICE_COMPAT_VERSION := 1.1
> > DEVICE_COMPAT_MESSAGE := Config cannot be migrated from swconfig to DSA
> > @@ -1960,7 +1977,7 @@ define Device/ubnt_unifi-6-lite
> > DEVICE_MODEL := UniFi 6 Lite
> > DEVICE_DTS_CONFIG := config at 1
> > DEVICE_PACKAGES += kmod-mt7603 kmod-mt7915e
> > - KERNEL := kernel-bin | lzma | fit lzma $$(KDIR)/image-$$(firstword $$(DEVICE_DTS)).dtb
> > + KERNEL := kernel-bin | lzma | fit-dtalign lzma $$(KDIR)/image-$$(firstword $$(DEVICE_DTS)).dtb
> > IMAGE_SIZE := 15424k
> > endef
> > TARGET_DEVICES += ubnt_unifi-6-lite
> > --
> > 2.30.2
> >
> >
> > _______________________________________________
> > openwrt-devel mailing list
> > openwrt-devel at lists.openwrt.org
> > https://lists.openwrt.org/mailman/listinfo/openwrt-devel
>
> --
> Ansuel
>
> _______________________________________________
> openwrt-devel mailing list
> openwrt-devel at lists.openwrt.org
> https://lists.openwrt.org/mailman/listinfo/openwrt-devel
More information about the openwrt-devel
mailing list