Bug #3015: NetBSD vs test/dl

• Status: Closed
• Assignee: tenderlovemaking (Aaron Patterson)
• Target version: 2.0.0
• ruby -v: ruby 1.9.2dev (2010-03-26 trunk 27055) [i386-netbsdelf5.0.]

[skandragon (Michael Graff)]

=begin
There are probably related problems with the mentioned tests, or the code they are testing, on NetBSD 5.0.2.

I have not yet investigated the cause of these.

1. Error:
   test_to_ptr_io(DL::TestCPtr):
   RangeError: bignum too big to convert into long' /u1/home/explorer/proj/ruby/test/dl/test_cptr.rb:84:in call'
   /u1/home/explorer/proj/ruby/test/dl/test_cptr.rb:84:in block in test_to_ptr_io' /u1/home/explorer/proj/ruby/test/dl/test_cptr.rb:81:in open'
   /u1/home/explorer/proj/ruby/test/dl/test_cptr.rb:81:in `test_to_ptr_io'

2. Error:
   test_callback(DL::TestDL):
   RangeError: bignum too big to convert into long' /u1/home/explorer/proj/ruby/test/dl/test_dl2.rb:143:in call'
   /u1/home/explorer/proj/ruby/test/dl/test_dl2.rb:143:in `test_callback'

3. Error:
   test_strcpy(DL::TestDL):
   RangeError: bignum too big to convert into long' /u1/home/explorer/proj/ruby/test/dl/test_dl2.rb:134:in call'
   /u1/home/explorer/proj/ruby/test/dl/test_dl2.rb:134:in `test_strcpy'

4. Error:
   test_static_sym(DL::TestHandle):
   DL::DLError: unknown symbol "dlopen"
   /u1/home/explorer/proj/ruby/test/dl/test_handle.rb:26:in sym' /u1/home/explorer/proj/ruby/test/dl/test_handle.rb:26:in test_static_sym'
   =end

[tenderlovemaking (Aaron Patterson)]

• Category set to ext
• Assignee set to tenderlovemaking (Aaron Patterson)
• Target version set to 1.9.2

=begin

=end

[skandragon (Michael Graff)]

=begin
On 3/25/10 8:22 PM, Aaron Patterson wrote:

Most likely this will be fixed when I get the dl/ffi underpinnings
merged to trunk. In the mean time, could you try against my ruby fork
that has ffi integrated?

https://github.com/tenderlove/ruby/tree/ffi

Certainly! I do not know what ffi is, or how it will help me, but I
will shortly :)

Cloning now.

--Michael

=end

[skandragon (Michael Graff)]

=begin
On 3/25/10 8:22 PM, Aaron Patterson wrote:

Most likely this will be fixed when I get the dl/ffi underpinnings
merged to trunk. In the mean time, could you try against my ruby fork
that has ffi integrated?

Trying that gives a different set of errors mostly:

1. Error:
   test_to_ptr_io(DL::TestCPtr):
   RangeError: bignum too big to convert into long' /u1/home/explorer/proj/ruby-ffi/ruby/test/dl/test_cptr.rb:84:in call'
   /u1/home/explorer/proj/ruby-ffi/ruby/test/dl/test_cptr.rb:84:in
   block in test_to_ptr_io' /u1/home/explorer/proj/ruby-ffi/ruby/test/dl/test_cptr.rb:81:in open'
   /u1/home/explorer/proj/ruby-ffi/ruby/test/dl/test_cptr.rb:81:in
   `test_to_ptr_io'

2. Error:
   test_NEXT(DL::TestHandle):
   DL::DLError: unknown symbol "malloc"
   /u1/home/explorer/proj/ruby-ffi/ruby/test/dl/test_handle.rb:128:in []' /u1/home/explorer/proj/ruby-ffi/ruby/test/dl/test_handle.rb:128:in test_NEXT'

3. Error:
   test_static_sym(DL::TestHandle):
   DL::DLError: unknown symbol "dlopen"
   /u1/home/explorer/proj/ruby-ffi/ruby/test/dl/test_handle.rb:25:in sym' /u1/home/explorer/proj/ruby-ffi/ruby/test/dl/test_handle.rb:25:in test_static_sym'

I'm willing to give out a NetBSD account to you if that helps :)

--Michael

=end

[skandragon (Michael Graff)]

=begin
On 3/26/10 8:16 AM, Aaron Patterson wrote:

Are you on 32bit or 64bit? I'll try installing a virtual machine.

That machine is i386, so 32-bit. I have another VM there I can spin up
(and would have to update to latest code) that is 64 too.

sizeof(long) is certainly going to be 32-bit. Most of NetBSD system
calls are 64-bit for file offsets and the like, however. I'm not
certain if that matters much. Pointers are certainly 32-bit here.

I'm willing to give out a NetBSD account to you if that helps :)

If I can't get my virtual machine running, I'll take you up on that
offer. Thanks!

I have production-like machines that run NetBSD-5.0.x/i386 (32-bit)
and have VMs that run:

NetBSD-5.0.x/i386
NetBSD-5.0.x/amd64
NetBSD-current/i386
NetBSD-current/amd64

In general I don't mind sharing these with you if that will help make
NetBSD more supported. If there is some sort of automated build/test
environment I could run and contribute feedback to, I would set that up
as well on all four VMs.

My goal is to get NetBSD to be a first-class Ruby citizen. :)

--Michael

=end

[skandragon (Michael Graff)]

=begin
Has any progress been made on this that I can test?
=end

[mame (Yusuke Endoh)]

• Priority changed from Normal to 3
• Target version changed from 1.9.2 to 2.0.0

=begin
I set this ticket to Low priority because there is no maintainer
for NetBSD. A patch is welcome.

--
Yusuke Endoh [e-mail]
=end

[naruse (Yui NARUSE)]

• Status changed from Open to Closed

=begin
Fixed by r27890 and r27891, this also backported to 1.9.2.
=end