Bug #17792: make notes and make test fail with Ruby3.0.1p64 RaspberryPI 4B Ubuntu 20.10 ARM64

• Status: Open
• Assignee: -
• Target version: -
• ruby -v: ruby 3.0.1p64 (2021-04-05 revision 0fb782ee38) [aarch64-linux]
• Backport: 2.6: UNKNOWN, 2.7: UNKNOWN, 3.0: UNKNOWN

[hanlyusarang (Hanlyu Sarang)]

I am building Ruby 3.01 from sources on a RaspberryPI 4B running Ubuntu 20.10 ARM64.

This is my first day using this PI4B + Ubuntu 20.10 ARM64, and this is the first time I have attempted to build Ruby on it.
I received a few notes during compilation and one failure during testing.

Note that on the previous day, I built the same Ruby sources on the same machine running RaspberryPI OS (32-bit Debian), no problems. FYI, you can use either a 32-bit or a 64-bit OS on the RasberryPI 4B.

Anyway, here are the details of what I experienced on the PI4B using Ubuntu 20.10 ARM64:

During the make, I get a few notes:

compiling parse.c
parse.y: In function ‘node_assign’:
parse.y:11265:1: note: parameter passing for argument of type ‘struct lex_context’ changed in GCC 9.1
11265 | node_assign(struct parser_params *p, NODE *lhs, NODE *rhs, struct lex_context ctxt, const YYLTYPE *loc)
| ^~~~~~~~~~~
parse.y: In function ‘new_const_op_assign’:
parse.y:12283:1: note: parameter passing for argument of type ‘struct lex_context’ changed in GCC 9.1
12283 | new_const_op_assign(struct parser_params *p, NODE *lhs, ID op, NODE *rhs, struct lex_context ctxt, const YYLTYPE *loc)
| ^~~~~~~~~~~~~~~~~~~
parse.y: In function ‘new_op_assign’:
parse.y:12196:1: note: parameter passing for argument of type ‘struct lex_context’ changed in GCC 9.1
12196 | new_op_assign(struct parser_params *p, NODE *lhs, ID op, NODE *rhs, struct lex_context ctxt, const YYLTYPE *loc)
| ^~~~~~~~~~~~~
compiling proc.c

The build is successful, but when I run make test, I get one failure

/home/pi/builds/ruby-3.0.1# make test
BASERUBY = /usr/bin/ruby --disable=gems
CC = gcc
LD = ld
LDSHARED = gcc -shared
CFLAGS = -O3 -ggdb3 -Wall -Wextra -Wdeprecated-declarations -Wduplicated-cond -Wimplicit-function-declaration -Wimplicit-int -Wmisleading-indentation -Wpointer-arith -Wwrite-strings -Wimplicit-fallthrough=0 -Wmissing-noreturn -Wno-cast-function-type -Wno-constant-logical-operand -Wno-long-long -Wno-missing-field-initializers -Wno-overlength-strings -Wno-packed-bitfield-compat -Wno-parentheses-equality -Wno-self-assign -Wno-tautological-compare -Wno-unused-parameter -Wno-unused-value -Wsuggest-attribute=format -Wsuggest-attribute=noreturn -Wunused-variable -std=gnu99
XCFLAGS = -D_FORTIFY_SOURCE=2 -fstack-protector-strong -fno-strict-overflow -fvisibility=hidden -fexcess-precision=standard -DRUBY_EXPORT -fPIE -I. -I.ext/include/aarch64-linux -I./include -I. -I./enc/unicode/12.1.0
CPPFLAGS =
DLDFLAGS = -Wl,--compress-debug-sections=zlib -fstack-protector-strong -pie
SOLIBS = -lz -lpthread -lrt -lrt -lgmp -ldl -lcrypt -lm
LANG = en_US.UTF-8
LC_ALL =
LC_CTYPE =
MFLAGS =
gcc (Ubuntu 10.2.0-13ubuntu1) 10.2.0
Copyright (C) 2020 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

./revision.h unchanged
#655 test_io.rb:87:in `block in <top (required)>':
at_exit { p :foo }

   megacontent = "abc" * 12345678
   #File.open("megasrc", "w") {|f| f << megacontent }

   t0 = Thread.main
   Thread.new { sleep 0.001 until t0.stop?; Process.kill(:INT, $$) }

   r1, w1 = IO.pipe
   r2, w2 = IO.pipe
   t1 = Thread.new { w1 << megacontent; w1.close }
   t2 = Thread.new { r2.read; r2.close }
   IO.copy_stream(r1, w2) rescue nil
   w2.close
   r1.close
   t1.join
   t2.join

#=> killed by SIGKILL (signal 9) (timeout) megacontent-copy_stream
test_io.rb FAIL 1/9
Fiber count: 10000 (skipping)
FAIL 1/1486 tests failed
make: *** [uncommon.mk:766: yes-btest-ruby] Error 1

I went ahead and did the make install, which succeeded.
I now have an installation of Ruby which I will be working with over the next few months to see if it works OK.

If you need more info, please let me know.

[mame (Yusuke Endoh)]

Sorry for a less informative reply, but Ruby is not yet tested well with 64 bit Raspberry Pi. I have no environment to reproduce the issue. Contribution is welcome.

I see a similar random failure on our CI (raspbian10-aarch64, still beta) but I'm unsure if it is the same issue. And I cannot log in to the machine. @znz (Kazuhiro NISHIYAMA) can you check it out?
http://rubyci.s3.amazonaws.com/raspbian10-aarch64/ruby-master/log/20210410T023158Z.fail.html.gz

[hanlyusarang (Hanlyu Sarang)]

Another FYI: The RaspberryPI 4B uses a microSD card (class 10) for the OS and all installed software. I was building Ruby on a 32GB version of this card.

The I/O on a microSD card is much slower than on a 5400 RPM HD, and much, much slower than on an SSD. I know from experience that the speed difference between the microSD card and the SSD is something on the order of 10 for intensive I/O operations.

If the test routine that failed was using a timeout mechanism. then perhaps the amount of time allowed was insufficient. But when I build Ruby on the same machine and type of card using a 32-bit Debian-like OS, no test failures. So, maybe the media speed is not a factor?

[xtkoba (Tee KOBAYASHI)]

The test code at bootstraptest/test_io.rb:87 is as follows. Try running it separately and see what happens.

at_exit { p :foo }

megacontent = "abc" * 12345678
#File.open("megasrc", "w") {|f| f << megacontent }

t0 = Thread.main
Thread.new { sleep 0.001 until t0.stop?; Process.kill(:INT, $$) }

r1, w1 = IO.pipe
r2, w2 = IO.pipe
t1 = Thread.new { w1 << megacontent; w1.close }
t2 = Thread.new { r2.read; r2.close }
IO.copy_stream(r1, w2) rescue nil
w2.close
r1.close
t1.join
t2.join

Timeout does not tell whether the code execution just takes longer than expected or hangs up forever.

[hanlyusarang (Hanlyu Sarang)]

The test code took almost two minutes to run, and it failed:

# TEST RESULTS:
pi> ruby test.rb 
:foo
test.rb:13:in `copy_stream': Interrupt
	from test.rb:13:in `<main>'

# LINE 13: IO.copy_stream(r1, w2) rescue nil

# TEST CODE:
at_exit { p :foo }

megacontent = "abc" * 12345678
#File.open("megasrc", "w") {|f| f << megacontent }

t0 = Thread.main
Thread.new { sleep 0.001 until t0.stop?; Process.kill(:INT, $$) }

r1, w1 = IO.pipe
r2, w2 = IO.pipe
t1 = Thread.new { w1 << megacontent; w1.close }
t2 = Thread.new { r2.read; r2.close }
IO.copy_stream(r1, w2) rescue nil
w2.close
r1.close
t1.join
t2.join

[xtkoba (Tee KOBAYASHI)]

Interruption in copy_stream is what this test is intended for, and so the result of the test in #note-4 is a success. But taking about 2 minutes to finish is unexpected to me. Possibly due to a bug in the pthread library?

FWIW, the megacontent-copy_stream test in bootstraptest/test_io.rb is known to fail on Cygwin (#1388, #5055, #7781; all rejected). This is different from the case here in that it never finishes on Cygwin (which I have confirmed).

I also have found that the timeout issue of this test is reported for kFreeBSD (https://bugs.debian.org/881779).