Re: [Linux-ia64] toolchain status

Re: [Linux-ia64] toolchain status

[Francis Galiegue]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 935 bytes --]

On Thu, 16 Nov 2000, H . J . Lu wrote:

> > > There is no DT_INIT_ARRAY in libc.so. Even if there is DT_INIT_ARRAY,
> > > I still don't know how to use it to tell where DT_INIT points to.
> > 
> > Easy enough -- you know from DT_PLTGOT what the GP is for the DSO.
> > If *(long *)(DT_INIT + 8) == GP, then you can assume that DT_INIT
> > is a descriptor instead of an offset.
> > 
> 
> Thanks. This patch seems to work for me.
> 

I don't know which version this patch is supposed to apply to, but it doesn't
apply as is on the stock FSF tree. "Corrected" patch for this tree is attached,
in case it is of interest for someone.

Now if only I understood what you were talking about :)

-- 
Francis Galiegue, fg@mandrakesoft.com
"Programming is a race between programmers, who try and make more and more
idiot-proof software, and universe, which produces more and more remarkable
idiots. Until now, universe leads the race"  -- R. Cook

[-- Attachment #2: Type: TEXT/PLAIN, Size: 792 bytes --]

diff -urN glibc-2.2.old/sysdeps/ia64/dl-lookupcfg.h glibc-2.2/sysdeps/ia64/dl-lookupcfg.h
--- glibc-2.2.old/sysdeps/ia64/dl-lookupcfg.h	Thu Nov 16 18:55:16 2000
+++ glibc-2.2/sysdeps/ia64/dl-lookupcfg.h	Thu Nov 16 19:03:46 2000
@@ -39,5 +39,13 @@
 					ElfW(Addr) start);
 
 #define DL_FUNCTION_ADDRESS(map, addr) _dl_function_address ((map), (addr))
+#if 0
 #define DL_DT_INIT_ADDRESS(map, addr) DL_FUNCTION_ADDRESS ((map), (addr))
 #define DL_DT_FINI_ADDRESS(map, addr) DL_FUNCTION_ADDRESS ((map), (addr))
+#else
+#define DL_DT_INIT_ADDRESS(map, addr)					      \
+  ((map)->l_info[DT_PLTGOT]->d_un.d_ptr == *(Elf64_Addr *) ((addr) + 8)       \
+   ? (addr) : DL_FUNCTION_ADDRESS (map, addr))
+#define DL_DT_FINI_ADDRESS(map, addr) DL_DT_INIT_ADDRESS (map, addr)
+#endif
+

Re: [Linux-ia64] toolchain status

[H . J . Lu]

On Fri, Nov 17, 2000 at 03:55:30PM +0100, Francis Galiegue wrote:
> On Thu, 16 Nov 2000, H . J . Lu wrote:
> 
> > > > There is no DT_INIT_ARRAY in libc.so. Even if there is DT_INIT_ARRAY,
> > > > I still don't know how to use it to tell where DT_INIT points to.
> > > 
> > > Easy enough -- you know from DT_PLTGOT what the GP is for the DSO.
> > > If *(long *)(DT_INIT + 8) = GP, then you can assume that DT_INIT
> > > is a descriptor instead of an offset.
> > > 
> > 
> > Thanks. This patch seems to work for me.
> > 
> 
> I don't know which version this patch is supposed to apply to, but it doesn't
> apply as is on the stock FSF tree. "Corrected" patch for this tree is attached,
> in case it is of interest for someone.

Your patch looks wrong. Please check out the glibc 2.2 from CVS again.


-- 
H.J. Lu (hjl@valinux.com)

Re: [Linux-ia64] toolchain status

[H . J . Lu]

On Fri, Nov 17, 2000 at 06:29:26PM +0100, Francis Galiegue wrote:
> On Fri, 17 Nov 2000, H . J . Lu wrote:
> 
> > > 
> > > I don't know which version this patch is supposed to apply to, but it doesn't
> > > apply as is on the stock FSF tree. "Corrected" patch for this tree is attached,
> > > in case it is of interest for someone.
> > 
> > Your patch looks wrong. Please check out the glibc 2.2 from CVS again.
> > 
> 
> I'm talking about GNU's stock glibc-2.2.tar.gz here... I didn't know there
> were CVS snapshots of it again. Where are they to be found?
> 

http://sources.redhat.com/glibc/

-- 
H.J. Lu (hjl@valinux.com)

Re: [Linux-ia64] toolchain status

[Francis Galiegue]

On Fri, 17 Nov 2000, H . J . Lu wrote:

> > 
> > I don't know which version this patch is supposed to apply to, but it doesn't
> > apply as is on the stock FSF tree. "Corrected" patch for this tree is attached,
> > in case it is of interest for someone.
> 
> Your patch looks wrong. Please check out the glibc 2.2 from CVS again.
> 

I'm talking about GNU's stock glibc-2.2.tar.gz here... I didn't know there
were CVS snapshots of it again. Where are they to be found?

-- 
Francis Galiegue, fg@mandrakesoft.com
"Programming is a race between programmers, who try and make more and more
idiot-proof software, and universe, which produces more and more remarkable
idiots. Until now, universe leads the race"  -- R. Cook

Re: [Linux-ia64] toolchain status

[Francis Galiegue]

On Fri, 17 Nov 2000, H . J . Lu wrote:

> > 
> > I'm talking about GNU's stock glibc-2.2.tar.gz here... I didn't know there
> > were CVS snapshots of it again. Where are they to be found?
> > 
> 
> http://sources.redhat.com/glibc/
> 

Hmmm, in snapshots/, the most recent I see is 2.1.3-pre4. This obviously is not
the good one...

-- 
Francis Galiegue, fg@mandrakesoft.com
"Programming is a race between programmers, who try and make more and more
idiot-proof software, and universe, which produces more and more remarkable
idiots. Until now, universe leads the race"  -- R. Cook

Re: [Linux-ia64] toolchain status

[Dan Pop]

On Mon, 20 Nov 2000, Francis Galiegue wrote:

> On Fri, 17 Nov 2000, H . J . Lu wrote:
> 
> > > 
> > > I'm talking about GNU's stock glibc-2.2.tar.gz here... I didn't know there
> > > were CVS snapshots of it again. Where are they to be found?
> > > 
> > 
> > http://sources.redhat.com/glibc/
> > 
> 
> Hmmm, in snapshots/, the most recent I see is 2.1.3-pre4. This obviously is not
> the good one...

You've missed these bits from the URL above:

    You can access the development source tree a couple of different ways. 

    Anonymous CVS read-only access 

	 cvs -z 9 -d :pserver:anoncvs@anoncvs.cygnus.com:/cvs/glibc login
	 {enter "anoncvs" as the password}
	 cvs -z 9 -d :pserver:anoncvs@anoncvs.cygnus.com:/cvs/glibc co libc
	   

    Read-only web-based CVS access 
	 You can use the cvsweb interface. 

Dan