public inbox for gdb-patches@sourceware.org
 help / color / mirror / Atom feed
* RFA: m32r sim: Add prototypes for functions that pass/return DI values
@ 2020-11-10 17:42 Nick Clifton
  2020-11-10 18:26 ` Simon Marchi
  2020-11-11 21:13 ` Andrew Burgess
  0 siblings, 2 replies; 5+ messages in thread
From: Nick Clifton @ 2020-11-10 17:42 UTC (permalink / raw)
  To: dje, andrew.burgess; +Cc: law, gdb-patches

Hi Doug, Hi Andrew,

  Jeff Law has tracked down a set of problems with the m32r simulator
  which turned out to be because the m32rbf_h_accum_get_handler()
  function returns a DI value, but it was not prototyped.  So callers
  would assume that it returned an int, and happiness ensued...

  I am proposing the attached patch as a workaround for the problem.  A
  proper fix would be to update the scheme files used to generate the
  cpu headers.  But I am not a scheme programmer, and besides getting
  cgen to work with modern versions of guile seems to be quite
  difficult.  So instead the patch adds prototypes into one of the m32r
  sim header files that is not auto-generated.

  OK to apply ?

Cheers
  Nick

sim/m32r/ChangeLog
2020-11-10  Nick Clifton  <nickc@redhat.com>

	* m32r-sim.h (m32rbf_h_accum_get_handler): Always provide a
	prototype for this function.
	(m32rbf_h_accum_set_handler): Likewise.
	(m32r2f_h_accums_get_handler): Prototype.
	(m32r2f_h_accums_set_handler): Prototype.

diff --git a/sim/m32r/m32r-sim.h b/sim/m32r/m32r-sim.h
index 9f3e270f39..6f9f4610c0 100644
--- a/sim/m32r/m32r-sim.h
+++ b/sim/m32r/m32r-sim.h
@@ -61,10 +61,19 @@ extern void m32rbf_h_psw_set_handler (SIM_CPU *, UQI);
   XCONCAT2 (WANT_CPU,_h_psw_set_handler) (current_cpu, (val))
 #endif
 
-#ifndef  GET_H_ACCUM
+/* FIXME: These prototypes are necessary because the cgen generated
+   cpu.h, cpux.h and cpu2.h headers do not provide them, and functions
+   which take or return parameters that are larger than an int must be
+   prototyed in order for them to work correctly.
+
+   The correct solution is to fix the code in cgen/sim.scm to generate
+   prototypes for each of the functions it generates.  */
 extern DI   m32rbf_h_accum_get_handler (SIM_CPU *);
 extern void m32rbf_h_accum_set_handler (SIM_CPU *, DI);
+extern DI   m32r2f_h_accums_get_handler (SIM_CPU *, UINT);
+extern void m32r2f_h_accums_set_handler (SIM_CPU *, UINT, DI);
 
+#ifndef  GET_H_ACCUM
 #define GET_H_ACCUM() \
   XCONCAT2 (WANT_CPU,_h_accum_get_handler) (current_cpu)
 #define SET_H_ACCUM(val) \


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: RFA: m32r sim: Add prototypes for functions that pass/return DI values
  2020-11-10 17:42 RFA: m32r sim: Add prototypes for functions that pass/return DI values Nick Clifton
@ 2020-11-10 18:26 ` Simon Marchi
  2020-11-11 15:33   ` Nick Clifton
  2020-11-11 21:13 ` Andrew Burgess
  1 sibling, 1 reply; 5+ messages in thread
From: Simon Marchi @ 2020-11-10 18:26 UTC (permalink / raw)
  To: Nick Clifton, dje, andrew.burgess; +Cc: gdb-patches, law

On 2020-11-10 12:42 p.m., Nick Clifton via Gdb-patches wrote:
> Hi Doug, Hi Andrew,
>
>   Jeff Law has tracked down a set of problems with the m32r simulator
>   which turned out to be because the m32rbf_h_accum_get_handler()
>   function returns a DI value, but it was not prototyped.  So callers
>   would assume that it returned an int, and happiness ensued...
>
>   I am proposing the attached patch as a workaround for the problem.  A
>   proper fix would be to update the scheme files used to generate the
>   cpu headers.  But I am not a scheme programmer, and besides getting
>   cgen to work with modern versions of guile seems to be quite
>   difficult.  So instead the patch adds prototypes into one of the m32r
>   sim header files that is not auto-generated.
>
>   OK to apply ?
>
> Cheers
>   Nick

This kind of problem bit us in the past in GDB too.  I'd suggest trying
to make the sim build system enable any or all of these warnings to
catch such problems.

  -Wmissing-declarations
  -Wmissing-prototypes
  -Wold-style-declaration
  -Wold-style-definition
  -Wstrict-prototypes

Simon

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: RFA: m32r sim: Add prototypes for functions that pass/return DI values
  2020-11-10 18:26 ` Simon Marchi
@ 2020-11-11 15:33   ` Nick Clifton
  2021-01-13  4:35     ` Mike Frysinger
  0 siblings, 1 reply; 5+ messages in thread
From: Nick Clifton @ 2020-11-11 15:33 UTC (permalink / raw)
  To: Simon Marchi, dje, andrew.burgess; +Cc: gdb-patches, law

Hi Simon,

> This kind of problem bit us in the past in GDB too.  I'd suggest trying
> to make the sim build system enable any or all of these warnings to
> catch such problems.
> 
>    -Wmissing-declarations
>    -Wmissing-prototypes
>    -Wold-style-declaration
>    -Wold-style-definition
>    -Wstrict-prototypes

I think that this would be an excellent idea.  Although I think
that it might cause a lot of work for the maintainers of cgen
based simulators.

Cheers
   Nick



^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: RFA: m32r sim: Add prototypes for functions that pass/return DI values
  2020-11-10 17:42 RFA: m32r sim: Add prototypes for functions that pass/return DI values Nick Clifton
  2020-11-10 18:26 ` Simon Marchi
@ 2020-11-11 21:13 ` Andrew Burgess
  1 sibling, 0 replies; 5+ messages in thread
From: Andrew Burgess @ 2020-11-11 21:13 UTC (permalink / raw)
  To: Nick Clifton; +Cc: dje, law, gdb-patches

* Nick Clifton <nickc@redhat.com> [2020-11-10 17:42:15 +0000]:

> Hi Doug, Hi Andrew,
> 
>   Jeff Law has tracked down a set of problems with the m32r simulator
>   which turned out to be because the m32rbf_h_accum_get_handler()
>   function returns a DI value, but it was not prototyped.  So callers
>   would assume that it returned an int, and happiness ensued...
> 
>   I am proposing the attached patch as a workaround for the problem.  A
>   proper fix would be to update the scheme files used to generate the
>   cpu headers.  But I am not a scheme programmer, and besides getting
>   cgen to work with modern versions of guile seems to be quite
>   difficult.  So instead the patch adds prototypes into one of the m32r
>   sim header files that is not auto-generated.
> 
>   OK to apply ?
> 
> Cheers
>   Nick
> 
> sim/m32r/ChangeLog
> 2020-11-10  Nick Clifton  <nickc@redhat.com>
> 
> 	* m32r-sim.h (m32rbf_h_accum_get_handler): Always provide a
> 	prototype for this function.
> 	(m32rbf_h_accum_set_handler): Likewise.
> 	(m32r2f_h_accums_get_handler): Prototype.
> 	(m32r2f_h_accums_set_handler): Prototype.
> 
> diff --git a/sim/m32r/m32r-sim.h b/sim/m32r/m32r-sim.h
> index 9f3e270f39..6f9f4610c0 100644
> --- a/sim/m32r/m32r-sim.h
> +++ b/sim/m32r/m32r-sim.h
> @@ -61,10 +61,19 @@ extern void m32rbf_h_psw_set_handler (SIM_CPU *, UQI);
>    XCONCAT2 (WANT_CPU,_h_psw_set_handler) (current_cpu, (val))
>  #endif
>  
> -#ifndef  GET_H_ACCUM
> +/* FIXME: These prototypes are necessary because the cgen generated
> +   cpu.h, cpux.h and cpu2.h headers do not provide them, and functions
> +   which take or return parameters that are larger than an int must be
> +   prototyed in order for them to work correctly.
> +
> +   The correct solution is to fix the code in cgen/sim.scm to generate
> +   prototypes for each of the functions it generates.  */
>  extern DI   m32rbf_h_accum_get_handler (SIM_CPU *);
>  extern void m32rbf_h_accum_set_handler (SIM_CPU *, DI);
> +extern DI   m32r2f_h_accums_get_handler (SIM_CPU *, UINT);
> +extern void m32r2f_h_accums_set_handler (SIM_CPU *, UINT, DI);

It looks like someone already did some of this before, so it seems
hard to argue against it :-)

Feel free to apply.

Thanks,
Andrew



>  
> +#ifndef  GET_H_ACCUM
>  #define GET_H_ACCUM() \
>    XCONCAT2 (WANT_CPU,_h_accum_get_handler) (current_cpu)
>  #define SET_H_ACCUM(val) \
> 

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: RFA: m32r sim: Add prototypes for functions that pass/return DI values
  2020-11-11 15:33   ` Nick Clifton
@ 2021-01-13  4:35     ` Mike Frysinger
  0 siblings, 0 replies; 5+ messages in thread
From: Mike Frysinger @ 2021-01-13  4:35 UTC (permalink / raw)
  To: Nick Clifton; +Cc: Simon Marchi, dje, andrew.burgess, gdb-patches, law

[-- Attachment #1: Type: text/plain, Size: 912 bytes --]

On 11 Nov 2020 15:33, Nick Clifton via Gdb-patches wrote:
> > This kind of problem bit us in the past in GDB too.  I'd suggest trying
> > to make the sim build system enable any or all of these warnings to
> > catch such problems.
> > 
> >    -Wmissing-declarations
> >    -Wmissing-prototypes
> >    -Wold-style-declaration
> >    -Wold-style-definition
> >    -Wstrict-prototypes
> 
> I think that this would be an excellent idea.  Although I think
> that it might cause a lot of work for the maintainers of cgen
> based simulators.

well i've turned on the full set of warnings (copied from gdb in days
past) for all sim ports now :).  when possible, i've turned on -Werror,
but you're right that all the cgen ports need a lot of love ...

i see gdb has split their warning code out into gdbsupport/.  i wonder
if we can reuse that, or maybe just sync the updated set of warnings.
-mike

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

^ permalink raw reply	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2021-01-13  4:36 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-11-10 17:42 RFA: m32r sim: Add prototypes for functions that pass/return DI values Nick Clifton
2020-11-10 18:26 ` Simon Marchi
2020-11-11 15:33   ` Nick Clifton
2021-01-13  4:35     ` Mike Frysinger
2020-11-11 21:13 ` Andrew Burgess

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).