On Thu, May 15, 2008 at 05:33:40PM -0600, Wesley J. Landaker wrote:
> Hi folks,
>
> I'm the Debian maintainer for GHDL, and I'm trying to build with gnat-4.3.
> For reference, it works fine with gnat-4.2, but gnat-4.3 is giving me a lot
> of trouble, i.e.:
There are two ways to get rid of this issue:
* do not use pragma no_run_time (what you did).
* switch to configurable run-time
- either by using a -gnatd? flag (I have to find which one)
- or by copying system.ads into the grt/ dir, modifying one or two
constant and compiling grt with this local system.ads
> Unfortunately, this does in fact build, but then trying to use the resulting
> tool gets me stuff like this when I try to use it:
>
> $ ghdl -m tb_blowfish
> analyze BlowfishPiROM.vhdl
> analyze BlowfishPArray.vhdl
> analyze BlowfishSBox.vhdl
> analyze BlowfishRound.vhdl
> analyze BlowfishCipher.vhdl
> analyze Blowfish.vhdl
> analyze tb_Blowfish.vhdl
> elaborate tb_blowfish
> $ ./tb_blowfish
> /usr/lib/ghdl/lib/gcc/i486-linux-gnu/4.1.2/vhdl/src/ieee/numeric_std-body.v93:2098:7:@20ns:
> (assertion warning): ./tb_blowfish:error: invalid memory access (dangling
> accesses or stack size too small)
> ./tb_blowfish:error: simulation failed
This is unexpected and shouldn't be related to not using the .adc file.
> Any ideas/thoughts? Shortly, gnat-4.2 is being removed from Debian, and if I
> can't resolve this, the GHDL package will probably have to do something
> somewhat obnoxious like bundle it's own copy of gnat.
Could you compile with -g and run within the debugger to see where this crash
happen ?
Did this code work with a previous ghdl compiler (I assume yes!)
Tristan.
_______________________________________________
Ghdl-discuss mailing list
[email protected]
https://mail.gna.org/listinfo/ghdl-discuss
