[RFC] sim/sh: Fix compile warnings for 64-bit build hosts

[RFC] sim/sh: Fix compile warnings for 64-bit build hosts

[Kevin Buettner]

Any comments on the following patch?

It fixes a bunch of warnings when building on a 64-bit host.  Here
are a few of them:

./code.c:93: warning: cast from pointer to integer of different size
./code.c:215: warning: cast from pointer to integer of different size
./code.c:230: warning: cast from pointer to integer of different size

Kevin

	* sh/interp.c (MA): Adjust cast to avoid warning on 64-bit hosts.

Index: interp.c
===================================================================
RCS file: /cvs/src/src/sim/sh/interp.c,v
retrieving revision 1.23
diff -u -p -r1.23 interp.c
--- interp.c	16 Apr 2011 18:16:36 -0000	1.23
+++ interp.c	16 Feb 2012 00:16:15 -0000
@@ -862,7 +862,7 @@ do { \
 #else
 
 #define MA(n) \
-  do { memstalls += ((((int) PC & 3) != 0) ? (n) : ((n) - 1)); } while (0)
+  do { memstalls += ((((long) PC & 3) != 0) ? (n) : ((n) - 1)); } while (0)
 
 #define L(x)   thislock = x;
 #define TL(x)  if ((x) == prevlock) stalls++;

Re: [RFC] sim/sh: Fix compile warnings for 64-bit build hosts

[Mike Frysinger]

[-- Attachment #1: Type: Text/Plain, Size: 594 bytes --]

On Wednesday 15 February 2012 19:26:22 Kevin Buettner wrote:
> Any comments on the following patch?

uintptr_t i think would be better, but we don't seem to use that in sim/, so 
this is OK

> It fixes a bunch of warnings when building on a 64-bit host.  Here
> are a few of them:
> 
> ./code.c:93: warning: cast from pointer to integer of different size
> ./code.c:215: warning: cast from pointer to integer of different size
> ./code.c:230: warning: cast from pointer to integer of different size

makes me think that this sim is incorrectly mixing target and host types
-mike

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [RFC] sim/sh: Fix compile warnings for 64-bit build hosts

[Kevin Buettner]

On Wed, 15 Feb 2012 23:50:34 -0500
Mike Frysinger <vapier@gentoo.org> wrote:

> On Wednesday 15 February 2012 19:26:22 Kevin Buettner wrote:
> > Any comments on the following patch?
> 
> uintptr_t i think would be better, but we don't seem to use that in sim/, so 
> this is OK
> 
> > It fixes a bunch of warnings when building on a 64-bit host.  Here
> > are a few of them:
> > 
> > ./code.c:93: warning: cast from pointer to integer of different size
> > ./code.c:215: warning: cast from pointer to integer of different size
> > ./code.c:230: warning: cast from pointer to integer of different size
> 
> makes me think that this sim is incorrectly mixing target and host types

My reading of the code shows that (the variable behind) PC is char *.
(PC is a macro defined to be insn_ptr.) The MA macro needs to convert
that pointer to an int type in order to apply the mask.  So I agree
that uintptr_t would be the right thing to use.

I've committed my patch, but if you wish I can try it with uintptr_t.
This will require us to include stdint.h.

Kevin

Re: [RFC] sim/sh: Fix compile warnings for 64-bit build hosts

[Mike Frysinger]

[-- Attachment #1: Type: Text/Plain, Size: 1491 bytes --]

On Thursday 16 February 2012 18:31:16 Kevin Buettner wrote:
> On Wed, 15 Feb 2012 23:50:34 -0500 Mike Frysinger wrote:
> > On Wednesday 15 February 2012 19:26:22 Kevin Buettner wrote:
> > > Any comments on the following patch?
> > 
> > uintptr_t i think would be better, but we don't seem to use that in sim/,
> > so this is OK
> > 
> > > It fixes a bunch of warnings when building on a 64-bit host.  Here
> > > are a few of them:
> > > 
> > > ./code.c:93: warning: cast from pointer to integer of different size
> > > ./code.c:215: warning: cast from pointer to integer of different size
> > > ./code.c:230: warning: cast from pointer to integer of different size
> > 
> > makes me think that this sim is incorrectly mixing target and host types
> 
> My reading of the code shows that (the variable behind) PC is char *.
> (PC is a macro defined to be insn_ptr.) The MA macro needs to convert
> that pointer to an int type in order to apply the mask.  So I agree
> that uintptr_t would be the right thing to use.
> 
> I've committed my patch, but if you wish I can try it with uintptr_t.
> This will require us to include stdint.h.

i don't think fixing this tiny issue would measurably improve the code, so 
don't worry about it.  to truly make the code better, the PC would need to not 
be a pointer but rather uint32_t (for 32bit SuperH procs) and uin64_t (for 
64bit SuperH procs).  and i imagine that is way more work than you want to 
take on ;).
-mike

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 836 bytes --]