bugGNUstep - Bugs: bug #34724, compiling gnustep-base with...

Group
 
 

bug #34724: compiling gnustep-base with libobjc2 and ffcall fails

Submitter:  Sebastian Reitenbach <buzzdee>
Submitted:  Wed 02 Nov 2011 09:58:11 AM UTC
   
 
Category:  Base/Foundation Severity:  3 - Normal
Item Group:  None Status:  Wont Fix
Privacy:  Public Assigned to:  None
Open/Closed:  Declined
* Mandatory Fields

Add a New Comment Rich Markup
   

Jump to the original submission

Sat 05 Nov 2011 06:22:11 PM UTC, comment #10: 

Yeah, I tried it only out of curiosity and to compare if I get other results with the test suite, knowing that use of ffcall is deprecated. Only opened it in case someone might be interested.

So close it if you like.

Sebastian Reitenbach <buzzdee>
Group Member
Sat 05 Nov 2011 05:58:28 PM UTC, comment #9: 

Is it possible to close this bug report now? The use of ffcall with libobjc2 was never intended and using ffi should be functional after Richards change.

Fred Kiefer <FredKiefer>
Group Member
Wed 02 Nov 2011 05:05:23 PM UTC, comment #8: 

I fixed a sparc64 specific bug passing small scalars via proxies when using libffi.
That makes most of the base tests pass for me.

Richard Frith-Macdonald <CaS>
Group Member
Wed 02 Nov 2011 04:38:35 PM UTC, comment #7: 

But I did find a solaris10 sparc64 machine to build and test on ...
It has gcc-4.1.1 and the libobjc that comes with it, and has libffi-3.0.5 installed.
I updated gnustep-make and gnustep-base from svn trunk.

I got similar invocation and proxy failures to yours.

Richard Frith-Macdonald <CaS>
Group Member
Wed 02 Nov 2011 03:21:05 PM UTC, comment #6: 

I tried to compile "old" gnustep libobjc 1.7.2 on sparc64 OpenBSD, but had to find out, it probably nobody tried before.

I had to

cp -r config/sparc config/sparc64 to allow it to find the tconfig.h file. It worked afterwards, but I'm still unsure whether this is the right thing to do...

Sebastian

Sebastian Reitenbach <buzzdee>
Group Member
Wed 02 Nov 2011 02:39:00 PM UTC, comment #5: 

Sorry ... I was wrong ... I checked the machine I thought was a solaris sparc64 system and found it was actually a solaris intel 64 bit system.


Richard Frith-Macdonald <CaS>
Group Member
Wed 02 Nov 2011 01:37:25 PM UTC, comment #4: 

which libobjc did you used? The "old" gnustep libobjc? or the libobjc that ships with the compiler, whatever the compiler is on the sparc solaris...

Sebastian

Sebastian Reitenbach <buzzdee>
Group Member
Wed 02 Nov 2011 11:15:47 AM UTC, comment #3: 

I've been away from home a lot the last few weeks, and hadn't picked up on that thread.

Looking at that email, it does look like most failures are with forwarding invocations (though some failures look like retain/release issues perhaps to do with threading).

I think all these tests passed recently on 64bit sparc solaris for me though.

Richard Frith-Macdonald <CaS>
Group Member
Wed 02 Nov 2011 10:28:14 AM UTC, comment #2: 

See the backtraces on -email is unavailable- in the initial mail I sent of the thread: failing -base tests on OpenBSD 5.0 -current on sparc64

this is with gcc and gcc libobjc2, I also tried with gcc and gcc libobjc, which had even more failing tests.

Therefore I just wanted to test with ffcall, just to compare whether there are differences, or libffi is broken.

Tested with libffi-3.0.9, and 3.0.10. With both versions, all libffi regression tests are passing. The output at the end is the same as on i386.

Sebastian

Sebastian Reitenbach <buzzdee>
Group Member
Wed 02 Nov 2011 10:16:28 AM UTC, comment #1: 

What's the problem with libffi?

I ask because ffcall is deprecated, and generally hasn't been used for a while, so I expect it has never even been tried, let alone worked, on any system with libobjc2 ... so there's probably a much better chance (less work) to get libffi working.

Richard Frith-Macdonald <CaS>
Group Member
Wed 02 Nov 2011 09:58:11 AM UTC, original submission:  

since the problems with libffi, or potential problems, I'm trying to compile gnustep-base with ffcall on sparc64 OpenBSD.

but it ends here:

GSFFCallInvocation.m: In function 'gs_method_for_receiver_and_selector':
GSFFCallInvocation.m:302: error: 'METHOD_NULL' undeclared (first use in this function)
GSFFCallInvocation.m:302: error: (Each undeclared identifier is reported only once
GSFFCallInvocation.m:302: error: for each function it appears in.)
GSFFCallInvocation.m:303: warning: control reaches end of non-void function
GSFFCallInvocation.m: In function 'gs_find_best_typed_sel':
GSFFCallInvocation.m:325: warning: 'sel_get_type' is deprecated (declared at /usr/local/include/objc/runtime-deprecated.h:67)
GSFFCallInvocation.m:331: warning: 'sel_get_any_typed_uid' is deprecated (declared at /usr/local/include/objc/runtime-deprecated.h:50)
GSFFCallInvocation.m:332: warning: 'sel_get_type' is deprecated (declared at /usr/local/include/objc/runtime-deprecated.h:67)
GSFFCallInvocation.m: In function 'gs_find_by_receiver_best_typed_sel':
GSFFCallInvocation.m:352: warning: 'sel_get_type' is deprecated (declared at /usr/local/include/objc/runtime-deprecated.h:67)
GSFFCallInvocation.m:369: error: dereferencing pointer to incomplete type
GSFFCallInvocation.m: In function 'gs_objc_msg_forward':
GSFFCallInvocation.m:459: warning: 'sel_get_type' is deprecated (declared at /usr/local/include/objc/runtime-deprecated.h:67)
GSFFCallInvocation.m: In function '+[GSFFCallInvocation load]':
GSFFCallInvocation.m:528: error: '__objc_msg_forward' undeclared (first use in this function)
GSFFCallInvocation.m: At top level:
GSFFCallInvocation.m:531: error: expected ')' before 'arglist_t'
GSFFCallInvocation.m: In function 'GSFFCallInvokeWithTargetAndImp':
GSFFCallInvocation.m:702: warning: value computed is not used
GSFFCallInvocation.m:702: warning: value computed is not used

will try with gcc libobjc too.

Sebastian Reitenbach <buzzdee>
Group Member

 

(Note: upload size limit is set to 16384 kB, after insertion of the required escape characters.)

Attach Files:
   
   
Comment:
   

No files currently attached

 

Depends on the following items: None found

Items that depend on this one: None found

 

Carbon-Copy List
  • -email is unavailable- added by FredKiefer (Posted a comment)
  • -email is unavailable- added by CaS (Posted a comment)
  • -email is unavailable- added by buzzdee (Submitted the item)
  •  

    There are 0 votes so far. Votes easily highlight which items people would like to see resolved in priority, independently of the priority of the item set by tracker managers.

    Only logged-in users can vote.

     

    Follow 2 latest changes.

    Date Changed by Updated Field Previous Value => Replaced by
    2011-11-07 FredKiefer StatusNone Wont Fix
        Open/ClosedOpen Declined

    Back to the top

    Powered by Savane 3.13-4448.
    Corresponding source code