Entware "Illegal instruction" running Pip packages after upgrade to 388.1 on RT-AX86U

[treefu]

Hello good people,

I noticed a jffs script failing (the script failed to run the idstools pip package actually) after upgrade to 388.1 and discovered I'm getting "Illegal instuction" when running any python3-pip package. It's been working for at least the last two years running pip packages happily.

I haven't tested extensively to find out which packages fail, but I've checked the USB disk pre-mount and found no e2fsck errors. It's mounted in the USB3 port (this router only has USB3 ports) and mounted as USB2 as per the Entware instructions.

I then also noticed pip itself not running:

admin@RT-AX86U-6580:/tmp/home/root# pip
Illegal instruction

So I tried uninstalling python3-pip:

admin@RT-AX86U-6580:/tmp/home/root# opkg remove python3-pip
Removing package python3-pip from root...

and installing again:

admin@RT-AX86U-6580:/tmp/home/root# opkg install python3-pip
Installing python3-pip (22.2.2-1) to root...
Downloading https://bin.entware.net/aarch64-k3.10/python3-pip_22.2.2-1_aarch64-3.10.ipk
Configuring python3-pip.

But then:

admin@RT-AX86U-6580:/tmp/home/root# pip install --upgrade pip
Illegal instruction

Kernel's not happy either:

Jan 19 21:24:47 kernel: potentially unexpected fatal signal 4.
Jan 19 21:24:47 kernel: CPU: 3 PID: 27253 Comm: pip Tainted: P           O    4.1.52 #2
Jan 19 21:24:47 kernel: Hardware name: Broadcom-v8A (DT)
Jan 19 21:24:47 kernel: task: ffffffc02f622040 ti: ffffffc021cf4000 task.ti: ffffffc021cf4000
Jan 19 21:24:47 kernel: PC is at 0x7fb31decc8
Jan 19 21:24:47 kernel: LR is at 0x7fb31d183c
Jan 19 21:24:47 kernel: pc : [<0000007fb31decc8>] lr : [<0000007fb31d183c>] pstate: 20000000
Jan 19 21:24:47 kernel: sp : 0000007fce043990
Jan 19 21:24:47 kernel: x29: 0000007fce043990 x28: 00000000112e0bf8
Jan 19 21:24:47 kernel: x27: 00000000112e0a70 x26: 00000000103c22fc
Jan 19 21:24:47 kernel: x25: 00000000112f18c0 x24: 0000007fb3f021a0
Jan 19 21:24:47 kernel: x23: 0000000000000000 x22: 0000007fb323c9a0
Jan 19 21:24:47 kernel: x21: 0000007fb3298258 x20: 0000007fb3335000
Jan 19 21:24:47 kernel: x19: 0000007fb3334000 x18: 000000001034fac0
Jan 19 21:24:47 kernel: x17: 0000000000000058 x16: 0000000011300de0
Jan 19 21:24:47 kernel: x15: 0000000000000000 x14: 0000000000000002
Jan 19 21:24:47 kernel: x13: 0000000000000060 x12: 000000001032c020
Jan 19 21:24:47 kernel: x11: 0000000000000000 x10: 0000000000000007
Jan 19 21:24:47 kernel: x9 : 000000000000270f x8 : 0000000000000062
Jan 19 21:24:47 kernel: x7 : 0000000000000000 x6 : 0000000000000001
Jan 19 21:24:47 kernel: x5 : 0000000000000881 x4 : 0000007fb3cd8000
Jan 19 21:24:47 kernel: x3 : 0000007fb41db010 x2 : 0000007fb333c4d4
Jan 19 21:24:47 kernel: x1 : 0000000000000000 x0 : 0000007fb3316fc0

I assume it's not the missing floating point processor issue others in the Internet have seen here, because

fp

is right there for the RT-AX86U:

admin@RT-AX86U-6580:/tmp/home/root# cat /proc/cpuinfo
processor       : 0
cpu model       : ARM
BogoMIPS        : 100.00
Features        : fp asimd evtstrm aes pmull sha1 sha2 crc32
CPU implementer : 0x42
CPU architecture: 8
CPU variant     : 0x0
CPU part        : 0x100
CPU revision    : 0


processor       : 1
cpu model       : ARM
BogoMIPS        : 100.00
Features        : fp asimd evtstrm aes pmull sha1 sha2 crc32
CPU implementer : 0x42
CPU architecture: 8
CPU variant     : 0x0
CPU part        : 0x100
CPU revision    : 0


processor       : 2
cpu model       : ARM
BogoMIPS        : 100.00
Features        : fp asimd evtstrm aes pmull sha1 sha2 crc32
CPU implementer : 0x42
CPU architecture: 8
CPU variant     : 0x0
CPU part        : 0x100
CPU revision    : 0


processor       : 3
cpu model       : ARM
BogoMIPS        : 100.00
Features        : fp asimd evtstrm aes pmull sha1 sha2 crc32
CPU implementer : 0x42
CPU architecture: 8
CPU variant     : 0x0
CPU part        : 0x100
CPU revision    : 0

I can downgrade to see if that helps, I've just not seen this issue before. Is there a trick I've missed anyone please?

Edit: Clarified Entware packages don't appear to be affected - just pip packages.

 

[treefu]

Okaay. It was a bad USB after all.

The problem didn't show up until I tried reformatting the USB device partitions (Optware, JFFS and Data), which gave I/O read errors.

USB Health check didn't detect it, but I wouldn't have expected it to.