[parisc-linux] My toolchain whoes...

[parisc-linux] My toolchain whoes...

[Matthew Hudson]

>From a gcc -E -dD

sigaction.c

In function __libc_sigaction

sigaction.c line 83:
      result = INLINE_SYSCALL (rt_sigaction, 4, sig,
                               act ? __ptrvalue (&kact) : NULL,
                               oact ? __ptrvalue (&koact) : NULL, _NSIG
/ 8);

gcc -E -dD output: (sorry I couldn't get this to you sooner willy I got
stuck in the middle of a bad day at work)
result = ({ unsigned long __sys_res; { register unsigned long __res
asm("r28"); register unsigned long __r24 __asm__("r24") = (unsigned
long)oact ? (&k_osigact) : ((void *)0); register unsigned long __r25
__asm__("r25") = (unsigned long)act ? (&k_sigact) : ((void *)0);
register unsigned long __r26__asm__("r26") = (unsigned long)sig; asm
volatile( "ble  0x100(%%sr2, %%r0)\n\t" " ldi %1, %%r20" : "=r" (__res)
: "i" ((__NR_sigaction)) , "r" (__r26), "r" (__r25), "r" (__r24) );
__sys_res = __res; } if (__sys_res >= (unsigned long)-4095) {
(*__errno_location ()) = (-__sys_res); __sys_res == (unsigned long)-1; }
__sys_res; });

undefined symbol: __NR_sigaction

Now, you tell me whats wrong?

My guess is that some how my config is getting mangled, I don't know
how.

Any ideas?

-Matt

Re: [parisc-linux] My toolchain whoes...

[Matthew Wilcox]

On Sat, Aug 26, 2000 at 06:49:27PM -0500, Matthew Hudson wrote:
> 
> >From a gcc -E -dD

I needed the whole file for that, not just the bit where it goes wrong.

Please send it to me off-list as it'll be a few hundred k, i bet.

-- 
Revolutions do not require corporate support.

Re: [parisc-linux] My toolchain whoes...

[David Huggins-Daines]

Matthew Hudson <mhudson@home.com> writes:

> gcc -E -dD output: (sorry I couldn't get this to you sooner willy I got
> stuck in the middle of a bad day at work)
> result = ({ unsigned long __sys_res; { register unsigned long __res
> asm("r28"); register unsigned long __r24 __asm__("r24") = (unsigned
> long)oact ? (&k_osigact) : ((void *)0); register unsigned long __r25
> __asm__("r25") = (unsigned long)act ? (&k_sigact) : ((void *)0);
> register unsigned long __r26__asm__("r26") = (unsigned long)sig; asm
> volatile( "ble  0x100(%%sr2, %%r0)\n\t" " ldi %1, %%r20" : "=r" (__res)
> : "i" ((__NR_sigaction)) , "r" (__r26), "r" (__r25), "r" (__r24) );
> __sys_res = __res; } if (__sys_res >= (unsigned long)-4095) {
> (*__errno_location ()) = (-__sys_res); __sys_res == (unsigned long)-1; }
> __sys_res; });
> 
> undefined symbol: __NR_sigaction
> 
> Now, you tell me whats wrong?

That's pretty screwed up, actually.  The line that says __NR_sigaction
clearly should say __NR_rt_sigaction (given that this is the output of
INLINE_SYSCALL(rt_sigaction, ...).  I don't know why it isn't.

Unless you pasted the wrong line ;-)

> My guess is that some how my config is getting mangled, I don't know
> how.

I'll say.  Maybe somewhere along the line some nasty header file did
something bad.  I'll see if the same thing is happening here.

-- 
dhd@linuxcare.com, http://www.linuxcare.com/
Linuxcare. Support for the revolution.