On 03/07/16 13:27, Walter F.J. Mueller wrote:
On 06/13/2016 08:22 PM, Tristan Gingold wrote:
On 12/06/16 17:25, Walter F.J. Mueller wrote:
Hi *,
I've some
.../ieee/numeric_std-body.v93:1005:7:@3418333332fs:(assertion
warning): NUMERIC_STD.">": metavalue detected, returning FALSE
warning and would like to find out the source code line which triggered
this.
Is there any mechanism to find out ?
Backtraces are available, except on x86-64 linux.
Dear Tristan,
I've installed ghdl 0.34dev on a 64 bit Intel system (thus
amd64 or x86-64), recompiled everything with -'-g', and
backtraces seem to work. So I don't understand your proviso
except on x86-64 linux
Ok, except on x86-64 linux when using mcode.
I get for example
tb_pdp11core --assert-level=warning
../../src/ieee/numeric_std-body.v93:1309:7:@0ms:(assertion warning):
NUMERIC_STD."<=": metavalue detected, returning FALSE
tb_pdp11core:error: assertion failed
from: ieee.numeric_std."<=" at numeric_std-body.v93:1314
from: process work.pdp11_vmbox(syn).proc_next at
pdp11_vmbox.vhd:372
tb_pdp11core:error: simulation failed
So far so good. However
- backtraces are only shown when simulator aborts
- no hierarchy info is included (unclear what current instance is)
So I wonder whether
- it is possible to print backtraces for all failed assertions, also
the ones which do not abort the simulation ? Should probably be enabled
via an option.
- it is possible to add the current process instance_name to the
backtrace info ? My current work around is to use --trace-processes
and look at the last 'run process' line before the backtrace.
With process instance name and vhdl stack one has the full information.
Yes, both are doable. Can you open an issue on github so that this idea
is not lost ?
Thanks,
Tristan.
_______________________________________________
Ghdl-discuss mailing list
Ghdl-discuss@gna.org
https://mail.gna.org/listinfo/ghdl-discuss
