[9fans] kl

[9fans] kl

[forsyth]

i think this is at least part of the problem (kl isn't up to date in
the released version, and hasn't got the xrefresolv changes).
it links cpu when i include them.  i suppose there was a backwards
dependency somewhere.

diff /n/dump/2002/0329/sys/src/cmd/kl/l.h ./l.h
224a225
> EXTERN	int	xrefresolv;
293c294
< void	loadlib(int, int);
---
> void	loadlib(void);
diff /n/dump/2002/0329/sys/src/cmd/kl/obj.c ./obj.c
16a17
>  *	-H3 -T0xE0004000 -R4		is javastation boot format
118a120,128
> 	case 3:	/* javastation boot */
> 		HEADR = 32L;
> 		if(INITTEXT == -1)
> 			INITTEXT = 0xE0004020L;
> 		if(INITDAT == -1)
> 			INITDAT = 0;
> 		if(INITRND == -1)
> 			INITRND = 4;
> 		break;
165c175
< 		loadlib(0, libraryp);
---
> 		loadlib();
195c205
< loadlib(int beg, int end)
---
> loadlib(void)
197c207,209
< 	int i, t;
---
> 	int i;
> 	long h;
> 	Sym *s;
199,200c211,213
< 	for(i=end-1; i>=beg; i--) {
< 		t = libraryp;
---
> loop:
> 	xrefresolv = 0;
> 	for(i=0; i<libraryp; i++) {
204,205d216
< 		if(t != libraryp)
< 			loadlib(t, libraryp);
206a218,222
> 	if(xrefresolv)
> 	for(h=0; h<nelem(hash); h++)
> 	for(s = hash[h]; s != S; s = s->link)
> 		if(s->type == SXREF)
> 			goto loop;
317a334
> 			xrefresolv = 1;

Re: [9fans] kl

[Lucio De Re]

On Tue, Apr 02, 2002 at 10:37:12AM +0100, forsyth@caldo.demon.co.uk wrote:
>
> i think this is at least part of the problem (kl isn't up to date in
> the released version, and hasn't got the xrefresolv changes).
> it links cpu when i include them.  i suppose there was a backwards
> dependency somewhere.
>
Validated.

I'm trying to build a "stable" system based on the March 27, 2001
distribution plus the few minor updates that followed.  That much
I have no trouble with.

More difficult is going through the mail since October 2001 and
picking all the little nits that have appeared on this list.  Maybe
in vain, I'd like to ask if anyone has actually tracked these
separately and can either point me to them or forward them to me.

Also, it may be asking a bit much, but perhaps we can add a short
recognisable keyword on the subject header when a patch is involved.
It would also help if one then stayed on topic <blush>.

++L