Re: [Gm2] Some More Snow Leopard Problems

[Gaius Mulley]

Michael Lambert <address@hidden> writes:

> Hi Gaius,
>
> I have some additional feedback on Snow Leopard (in addition to the
> patches previously submitted and the problem with the duplicate label
> generation in the assembly code).

Hi Michael,

many thanks - the patches are high on my do to list..

> For iso/run/pass/except2.mod, execution fails as follows.
> Unfortunately, the debugging isn't giving me line number information,
> even though this test is compiled with -g.  There are similar failures
> (KERN_PROTECTION_FAILURE) in except[48].mod.

ah interesting - when I run this on a i686 debian stable - it causes
the kernel to lock.. (I cannot remember which one actually causes
- but it appears when a make check-gm2 is performed.  Individually
the programs run fine, but collectively they will crash the kernel..
I need to explore this in more detail now..

> ----------
> kuru:testsuite lambert$ ./except2.x0
> keepFlying main body
> tryFlying main body
> fly main body
> terminate called after throwing an instance of 'int'
> Abort trap
> kuru:testsuite lambert$ gdb ./except2.x0
> GNU gdb 6.3.50-20050815 (Apple version gdb-1346) (Fri Sep 18 20:40:51
> UTC 2009)
> Copyright 2004 Free Software Foundation, Inc.
> GDB is free software, covered by the GNU General Public License, and
> you are
> welcome to change it and/or distribute copies of it under certain
> conditions.
> Type "show copying" to see the conditions.
> There is absolutely no warranty for GDB.  Type "show warranty" for
> details.
> This GDB was configured as "x86_64-apple-darwin"...Reading symbols for
> shared libraries .... done
>
> (gdb) run
> Starting program: /Volumes/Fantom_HFS+/gnu/gm2/obj/gcc/testsuite/
> except2.x0
> Reading symbols for shared libraries +++. done
> keepFlying main body
> tryFlying main body
> fly main body
>
> Program received signal EXC_BAD_ACCESS, Could not access memory.
> Reason: KERN_PROTECTION_FAILURE at address: 0x00000000
> 0x00013215 in fly ()
> (gdb) where
> #0  0x00013215 in fly ()
> #1  0x000132a4 in tryFlying ()
> #2  0x0001335b in keepFlying ()
> #3  0x000131a8 in _M2_except2_init ()
> #4  0x00001918 in init ()
> #5  0x00001a29 in main ()
> ----------

great thanks!  This is a super start

> As previously reported, compilation of iso/run/pass/long{,2,3}.mod
> fails with the duplicate assembler label.
>
> ----------
>
> Execution of iso/run/pass/shift[23].mod fails in a way which doesn't
> appear to help debugging:

ok, I've currently trying to fix ShiftLeft and family in SYSTEM.mod
I think there is a bug there and the test code just happened to work
due to previous stack values.


>
> ----------
> kuru:testsuite lambert$ ./shift2.x0
> kuru:testsuite lambert$ echo $?
> 1

the exit(1) is due to:

   b := large{1, 2, 3} ;
   b := SHIFT(b, 1) ;
   IF b#large{2, 3, 4}
   THEN
      exit(1)
   END ;

failing.  I've written a shift4.mod which attempts to shift 1 bit n
positions and it is highlighting a number of problems

> kuru:testsuite lambert$ gdb ./shift2.x0
> GNU gdb 6.3.50-20050815 (Apple version gdb-1346) (Fri Sep 18 20:40:51
> UTC 2009)
> Copyright 2004 Free Software Foundation, Inc.
> GDB is free software, covered by the GNU General Public License, and
> you are
> welcome to change it and/or distribute copies of it under certain
> conditions.
> Type "show copying" to see the conditions.
> There is absolutely no warranty for GDB.  Type "show warranty" for
> details.
> This GDB was configured as "x86_64-apple-darwin"...Reading symbols for
> shared libraries .... done

incidentally (off topic slightly) I've checked in a tiny gdb patch
which will allow gdb to read procedure types which have parameters of
the same procedure type.  If you need to debug iso test code, this is
very useful, as the iso libraries use this construct.  Without it gdb
crashes with out of memory

> (gdb) run
> Starting program: /Volumes/Fantom_HFS+/gnu/gm2/obj/gcc/testsuite/
> shift2.x0
> Reading symbols for shared libraries +++. done
>
> Program exited with code 01.
> ----------
>
> This one really seems strange:
>
> /Volumes/Fantom_HFS+/gnu/gm2/gcc-4.1.2/gcc/testsuite/gm2/iso/run/pass/
> testsystem.mod:111:4: error: total number of bit specified as
> parameters to MAKEADR exceeds address width
>
> If I'm not mistaken, the failure is with MAKEADR(0).

weird - yes!

> I still have some more errors to report, but this should be a start.

very true - most helpful thanks!

regards,
Gaius