CC @mundaym

Thanks for the bug report @richatbawag. A few requests:

1. Can you post the output of the command cat /proc/cpuinfo?
2. Is this a real machine or a qemu emulated machine?
3. Can you try again with the following environment variable set: GODEBUG="cpu.vx=off,cpu.vxe=off"?

The illegal instruction is pointing at an LGR instruction, but that is very unlikely to be accurate. Most likely the error is coming from the instruction before, a VLL instruction. We should only be trying to execute that if the vector facility is installed, but maybe the logic is slightly wrong.

1. Can you post the output of the command cat /proc/cpuinfo?

vendor_id       : IBM/S390
# processors    : 2
bogomips per cpu: 24038.00
max thread id   : 0
features	: esan3 zarch stfle msa ldisp eimm dfp etf3eh highgprs sie 
facilities      : 0 1 2 3 4 6 7 10 12 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 30 31 32 33 34 35 36 37 40 41 42 43 44 45 46 47 48 49 50 51 52 53 55 57 74 75 76 77 80 81 82 128 168
cache0          : level=1 type=Data scope=Private size=128K line_size=256 associativity=8
cache1          : level=1 type=Instruction scope=Private size=128K line_size=256 associativity=8
cache2          : level=2 type=Data scope=Private size=4096K line_size=256 associativity=8
cache3          : level=2 type=Instruction scope=Private size=4096K line_size=256 associativity=8
cache4          : level=3 type=Unified scope=Shared size=262144K line_size=256 associativity=32
cache5          : level=4 type=Unified scope=Shared size=983040K line_size=256 associativity=60
processor 0: version = FF,  identification = 0008F8,  machine = 2964
processor 1: version = FF,  identification = 0008F8,  machine = 2964

cpu number      : 0
cpu MHz dynamic : 5200
cpu MHz static  : 5200

cpu number      : 1
cpu MHz dynamic : 5200
cpu MHz static  : 5200

2. Is this a real machine or a qemu emulated machine?

A real machine, no qemu

3. Can you try again with the following environment variable set: GODEBUG="cpu.vx=off,cpu.vxe=off"?

Fails again.
go_1.15.2_build_form_source_tests_fail_s390x_cpuvxoff.txt

go1.14.9 works
go1.15 fails too

Thanks for the information. Looks like you are running a z13 with the vector facility disabled.

I think I know what the issue is. CL 156998 added calls to bytealg.IndexString that do not first check that the input string is shorter than bytealg.MaxLen (as required by the documentation for bytealg.IndexString). bytealg.MaxLen is 0 on s390x when the vector facility is unavailable and so bytealg.IndexString should not be called. Technically the generic implementation that was added in that CL should not be called either since generic platforms never set MaxLen to something other than 0.

Requiring callers to check MaxLen is probably asking for trouble. We might want to scrap it as part of the fix for this. I'll take a look and see if I can come up with a robust patch.

/cc @ianlancetaylor @randall77

Change https://golang.org/cl/256717 mentions this issue: bytes, internal/bytealg: fix incorrect IndexString usage

@gopherbot Please create backport issue for Go 1.15. This bug means that bytes.IndexAny and bytes.LastIndexAny cause an illegal instruction exception on s390x machines without the vector facility. There is no workaround.

Backport issue(s) opened: #41595 (for 1.15).

Remember to create the cherry-pick CL(s) as soon as the patch is submitted to master, according to https://golang.org/wiki/MinorReleases.