Project

General

Profile

Actions

Bug #18330

closed

Make failure on 32-bit Linux (Android) with Clang due to implicit 64-to-32-bit integer truncation

Added by xtkoba (Tee KOBAYASHI) 3 months ago. Updated about 1 month ago.

Status:
Closed
Priority:
Normal
Target version:
-
ruby -v:
ruby 3.1.0dev (2021-11-11 master 84202963c5) [arm-linux-android]
[ruby-core:106033]

Description

Observed in crossruby CI for armv7a-android30 (http://rubyci.s3.amazonaws.com/crossruby/crossruby-master-armv7a-android30/log/20211111T210030Z.log.html.gz):

compiling io_buffer.c
io_buffer.c:297:16: error: implicit conversion loses integer precision: 'off_t' (aka 'long long') to 'size_t' (aka 'unsigned int') [-Werror,-Wshorten-64-to-32]
        size = rb_file_size(io);
             ~ ^~~~~~~~~~~~~~~~
io_buffer.c:881:1: error: implicit conversion loses integer precision: 'uint64_t' (aka 'unsigned long long') to 'unsigned long' [-Werror,-Wshorten-64-to-32]
DECLAIR_TYPE(u64, uint64_t, RB_IO_BUFFER_LITTLE_ENDIAN, RB_ULONG2NUM, RB_NUM2ULONG, ruby_swap64)
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
io_buffer.c:855:17: note: expanded from macro 'DECLAIR_TYPE'
    return wrap(value); \
           ~~~~ ^~~~~
io_buffer.c:882:1: error: implicit conversion loses integer precision: 'uint64_t' (aka 'unsigned long long') to 'unsigned long' [-Werror,-Wshorten-64-to-32]
DECLAIR_TYPE(U64, uint64_t, RB_IO_BUFFER_BIG_ENDIAN, RB_ULONG2NUM, RB_NUM2ULONG, ruby_swap64)
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
io_buffer.c:855:17: note: expanded from macro 'DECLAIR_TYPE'
    return wrap(value); \
           ~~~~ ^~~~~
io_buffer.c:883:1: error: implicit conversion loses integer precision: 'int64_t' (aka 'long long') to 'long' [-Werror,-Wshorten-64-to-32]
DECLAIR_TYPE(s64, int64_t, RB_IO_BUFFER_LITTLE_ENDIAN, RB_LONG2NUM, RB_NUM2LONG, ruby_swap64)
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
io_buffer.c:855:17: note: expanded from macro 'DECLAIR_TYPE'
    return wrap(value); \
           ~~~~ ^~~~~
io_buffer.c:884:1: error: implicit conversion loses integer precision: 'int64_t' (aka 'long long') to 'long' [-Werror,-Wshorten-64-to-32]
DECLAIR_TYPE(S64, int64_t, RB_IO_BUFFER_BIG_ENDIAN, RB_LONG2NUM, RB_NUM2LONG, ruby_swap64)
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
io_buffer.c:855:17: note: expanded from macro 'DECLAIR_TYPE'
    return wrap(value); \
           ~~~~ ^~~~~
5 errors generated.
gmake: *** [Makefile:435: io_buffer.o] Error 1

Updated by jaruga (Jun Aruga) 3 months ago

For the arm32 + clang's log https://rubyci.s3.amazonaws.com/crossruby/crossruby-master-armv7a-android30/log/20211111T210030Z.log.html.gz.

If we want to add the arm32 + clang case to the pull-request timing CI, we can add the arm32 + clang case to the Cirrus CI.

Right now we have the arm32 + gcc case on Travis.
https://github.com/ruby/ruby/blob/415671a28273e5bfbe9aa00a0e386f025720ac23/.travis.yml#L122

We are using Cirrus CI for the Arm cases. Note Travis Arm pipeline is unstable for the more than 2 jobs.
https://github.com/ruby/ruby/blob/master/.cirrus.yml

Updated by Eregon (Benoit Daloze) 2 months ago

  • Assignee set to ioquatix (Samuel Williams)

Updated by ioquatix (Samuel Williams) 2 months ago

Thanks for this report.

Updated by ioquatix (Samuel Williams) 2 months ago

Does 32-bit Ruby support 64-bit wide integers? I guess no?

Updated by ioquatix (Samuel Williams) 2 months ago

https://github.com/ruby/ruby/pull/5114 should fix this, but do we have 32-bit CI to confirm?

Updated by jaruga (Jun Aruga) 2 months ago

Nice! Thanks!

Travis has the Arm 32-bit case, but not with clang but with gcc.
https://app.travis-ci.com/github/ruby/ruby/jobs/548818403

For RubyCI https://rubyci.org/ - the Raspbian 10 (arm) seems arm 32-bit. But I don't know how to login. Other servers in Ruby CI seems arm 64 bit.

Updated by ioquatix (Samuel Williams) 2 months ago

I merged this does it look like it’s fixed from your end? The RubyCI looks okay to me.

Updated by jaruga (Jun Aruga) 2 months ago

The RubyCI reported originally looks ok now.
crossruby-master-armv7a-android30
https://rubyci.s3.amazonaws.com/crossruby/crossruby-master-armv7a-android30/recent.html

Updated by ioquatix (Samuel Williams) about 1 month ago

  • Status changed from Open to Closed

Thanks!

Actions

Also available in: Atom PDF