Re: [PATCH] Consolidate compat_sys_waitid

[Matthew Wilcox <matthew@wil.cx>]

On Tue, Feb 15, 2005 at 02:01:49PM +1100, Stephen Rothwell wrote:
> Parisc seemed to assume th existance of compat_sys_waitid.

PA-RISC already has a compat_sys_waitid ;-P

The reason it isn't in Linus' tree yet is that it depends on the
is_compat_task() predicate which Andi vetoed out of Andrew's tree.
As a result, I haven't been able to merge any of the compat stuff
sitting in the PA tree.  A few more voices in favour of reintroducing
is_compat_task() would help.

> @@ -413,6 +414,36 @@ compat_sys_wait4(compat_pid_t pid, compa
>  	}
>  }
>  
> +asmlinkage long compat_sys_waitid(u32 which, u32 pid,
> +		struct compat_siginfo __user *uinfo, u32 options,
> +		struct compat_rusage __user *uru)

Some subtle differences which I feel incompetent to diagnose ... ours
looks like:

asmlinkage int compat_sys_waitid(int which, pid_t pid,
                                 compat_siginfo_t __user *infop, int options,
                                 struct compat_rusage __user *ru)


> +	BUG_ON(info.si_code & __SI_MASK);
> +	info.si_code |= __SI_CHLD;
> +	return copy_siginfo_to_user32(uinfo, &info);
> +}

Other than variable names, we're identical to this point:

        /* Tell copy_siginfo_to_user that it was __SI_CHLD */
        ksiginfo.si_code |= __SI_CHLD;
        
        if (compat_copy_siginfo_to_user(infop, &ksiginfo) != 0)
                return -EFAULT;

        return 0;
}

-- 
"Next the statesmen will invent cheap lies, putting the blame upon 
the nation that is attacked, and every man will be glad of those
conscience-soothing falsities, and will diligently study them, and refuse
to examine any refutations of them; and thus he will by and by convince 
himself that the war is just, and will thank God for the better sleep 
he enjoys after this process of grotesque self-deception." -- Mark Twain