[parisc-linux] Need five #defines, please.....

[parisc-linux] Need five #defines, please.....

[Christian Suder]

Ok,


I'm trying to skin the lizard (mozilla 0.9.7). I can build with the linux
default pthreads, but it doesn' work, windows show up but the content is
not drawn etc.. Hence I'd like to try a build with "classic" NSPR, which
implements it's own threading mechanism. Problem : There is a couple of
hardware-specific definitions that I don't know how to do, but somebody
with some insight in parisc-linux may be able to put it down in 5 minutes
(I hope..). Or maybe it has been solved it already ? 

Here's what I'm missing : 

#elif defined(__hppa__)
/* parisc */    
#define _MD_GET_SP(_t) 
#define _MD_SET_FP(_t, val)
#define _MD_GET_SP_PTR(_t) 
#define _MD_GET_FP_PTR(_t)
#define _MD_SP_TYPE 

As a hint, here some existing defines. As far as my limited incompetence
can tell these provide access to some of the context information of the
threads :

/* Sparc */  
#define _MD_GET_SP(_t) (_t)->md.context[0].__jmpbuf[JB_SP]
#define _MD_SET_FP(_t, val) ((_t)->md.context[0].__jmpbuf[JB_FP] = val)
#define _MD_GET_SP_PTR(_t) &(_MD_GET_SP(_t))
#define _MD_GET_FP_PTR(_t) (&(_t)->md.context[0].__jmpbuf[JB_FP])
#define _MD_SP_TYPE int

/* Alpha based Linux */
#define _MD_GET_SP(_t) (_t)->md.context[0].__jmpbuf[JB_SP]
#define _MD_SET_FP(_t, val)
#define _MD_GET_SP_PTR(_t) &(_MD_GET_SP(_t))
#define _MD_GET_FP_PTR(_t) ((void *) 0)
#define _MD_SP_TYPE long int


/* Intel based Linux */ 
#define _MD_GET_SP(_t) (_t)->md.context[0].__jmpbuf[JB_SP]
#define _MD_SET_FP(_t, val) ((_t)->md.context[0].__jmpbuf[JB_BP] = val)
#define _MD_GET_SP_PTR(_t) &(_MD_GET_SP(_t))                           
#define _MD_GET_FP_PTR(_t) (&(_t)->md.context[0].__jmpbuf[JB_BP])         
     #define _MD_SP_TYPE int


Thanks,
	Christian 

Re: [parisc-linux] Need five #defines, please.....

[kenneth westelinck]

Christian,

Can you keep me informed about this issue, as I am very keen on getting
mozilla to work on my 712/60. I've tried to compile the 0.9.5 version, a
long time ago, without any success (see earlier posts). I'm interested in
which files I need to modify, configure options, compiler options, ... .

Many thanks in advance.


regards,
Kenneth

----- Original Message -----
From: "Christian Suder" <csuder@cisco.com>
To: <parisc-linux@lists.parisc-linux.org>
Sent: Friday, January 04, 2002 10:25 AM
Subject: [parisc-linux] Need five #defines, please.....


> Ok,
>
>
> I'm trying to skin the lizard (mozilla 0.9.7). I can build with the linux
> default pthreads, but it doesn' work, windows show up but the content is
> not drawn etc.. Hence I'd like to try a build with "classic" NSPR, which
> implements it's own threading mechanism. Problem : There is a couple of
> hardware-specific definitions that I don't know how to do, but somebody
> with some insight in parisc-linux may be able to put it down in 5 minutes
> (I hope..). Or maybe it has been solved it already ?
>
> Here's what I'm missing :
>
> #elif defined(__hppa__)
> /* parisc */
> #define _MD_GET_SP(_t)
> #define _MD_SET_FP(_t, val)
> #define _MD_GET_SP_PTR(_t)
> #define _MD_GET_FP_PTR(_t)
> #define _MD_SP_TYPE
>
> As a hint, here some existing defines. As far as my limited incompetence
> can tell these provide access to some of the context information of the
> threads :
>
> /* Sparc */
> #define _MD_GET_SP(_t) (_t)->md.context[0].__jmpbuf[JB_SP]
> #define _MD_SET_FP(_t, val) ((_t)->md.context[0].__jmpbuf[JB_FP] = val)
> #define _MD_GET_SP_PTR(_t) &(_MD_GET_SP(_t))
> #define _MD_GET_FP_PTR(_t) (&(_t)->md.context[0].__jmpbuf[JB_FP])
> #define _MD_SP_TYPE int
>
> /* Alpha based Linux */
> #define _MD_GET_SP(_t) (_t)->md.context[0].__jmpbuf[JB_SP]
> #define _MD_SET_FP(_t, val)
> #define _MD_GET_SP_PTR(_t) &(_MD_GET_SP(_t))
> #define _MD_GET_FP_PTR(_t) ((void *) 0)
> #define _MD_SP_TYPE long int
>
>
> /* Intel based Linux */
> #define _MD_GET_SP(_t) (_t)->md.context[0].__jmpbuf[JB_SP]
> #define _MD_SET_FP(_t, val) ((_t)->md.context[0].__jmpbuf[JB_BP] = val)
> #define _MD_GET_SP_PTR(_t) &(_MD_GET_SP(_t))
> #define _MD_GET_FP_PTR(_t) (&(_t)->md.context[0].__jmpbuf[JB_BP])
>      #define _MD_SP_TYPE int
>
>
> Thanks,
> Christian
>
> _______________________________________________
> parisc-linux mailing list
> parisc-linux@lists.parisc-linux.org
> http://lists.parisc-linux.org/mailman/listinfo/parisc-linux
>

Re: [parisc-linux] Need five #defines, please.....

[Thomas Bogendoerfer]

On Fri, Jan 04, 2002 at 01:25:59AM -0800, Christian Suder wrote:
> I'm trying to skin the lizard (mozilla 0.9.7). I can build with the linux
> default pthreads, but it doesn' work, windows show up but the content is
> not drawn etc.. Hence I'd like to try a build with "classic" NSPR, which
> implements it's own threading mechanism. Problem : There is a couple of
> hardware-specific definitions that I don't know how to do, but somebody

I don't thing anyone has ported libc5 to hppa. Those missing defines are
libc5 hooks, iirc.

Fixing pthreads would be much more productive.

Thomas.

-- 
Crap can work. Given enough thrust pigs will fly, but it's not necessary a
good idea.                                 [ Alexander Viro on linux-kernel ]

Re: [parisc-linux] Need five #defines, please.....

[Christian Suder]

At 11:31 PM 1/4/2002 +0100, Thomas Bogendoerfer wrote:
>On Fri, Jan 04, 2002 at 01:25:59AM -0800, Christian Suder wrote:
> > I'm trying to skin the lizard (mozilla 0.9.7). I can build with the linux
> > default pthreads, but it doesn' work, windows show up but the content is
> > not drawn etc.. Hence I'd like to try a build with "classic" NSPR, which
> > implements it's own threading mechanism. Problem : There is a couple of
> > hardware-specific definitions that I don't know how to do, but somebody
>
>I don't thing anyone has ported libc5 to hppa. Those missing defines are
>libc5 hooks, iirc.

If I'm not mistaken libc5 <=> glibc1 and libc6 <=> glibc2, these are the 
ones for glibc2. In any case it runs on a (Linux-)Sparc with glibc2.


>Fixing pthreads would be much more productive.

Agree. However I believe the problems to be caused by sloppy programming in 
terms of thread handling, at least that was the issue I had with the pan 
newsreader (which behaved similar, btw). I don't think that can be fixed in 
the NSPR abstraction layer if it is caused by the application not calling 
the appropriate functions when needed. And I'm not going to dig through 160 
MB of source to find where somebody didn't initialize something 
correctly.... I'm going back and forth though to see if it is something hat 
can be done in NSPR.

Especially when in the end I don't want to run mozilla anyways. I just want 
the rendering engine to work so some more lightweight client like galeon 
can be used.

         Christian 

Re: [parisc-linux] Need five #defines, please.....

[Thomas Bogendoerfer]

On Fri, Jan 04, 2002 at 03:58:06PM -0800, Christian Suder wrote:
> If I'm not mistaken libc5 <=> glibc1 and libc6 <=> glibc2, these are the 
> ones for glibc2. In any case it runs on a (Linux-)Sparc with glibc2.

you are right, it's not a libc5 thing. I'm going to look at that tomorrow.

> >Fixing pthreads would be much more productive.
> 
> Agree. However I believe the problems to be caused by sloppy programming in 

right now I'm not sure, which mozilla gets further. Mine dies without 
display anything in nsGetModule of libimglib2.so. At least that's what 
gdb tells me. I've build my mozilla from the debian 0.9.7-2 sources.

Thomas.

-- 
Crap can work. Given enough thrust pigs will fly, but it's not necessary a
good idea.                                 [ Alexander Viro on linux-kernel ]