Re: ia32 sigwait patch

Re: ia32 sigwait patch

[Bjorn Helgaas]

On Monday 23 February 2004 11:52 am, you wrote:
> ia32 rt_sigtimewait() fails when some of the arguments are NULL. The
> fix is attached.

Sorry for the delay in responding; I've been out of the office for
a while.  I applied the patch (with the addition of a few spaces
in the ternary operator).

It looks to me like current 2.4 on x86_64 suffers from the same
problem with uts (sys_rt_sigtimedwait() gets a pointer to
uninitialized "t" when uts is NULL).

--- linux-2.4.21-4.EL/arch/ia64/ia32/ia32_signal.c.ori	2003-11-17 13:42:01.000000000 +0800
+++ linux-2.4.21-4.EL/arch/ia64/ia32/ia32_signal.c	2003-11-17 13:44:14.000000000 +0800
@@ -614,7 +614,7 @@
 			return -EFAULT;
 	}
 	set_fs(KERNEL_DS);
-	ret = sys_rt_sigtimedwait(&s, &info, &t, sigsetsize);
+	ret = sys_rt_sigtimedwait(&s, uinfo? &info :NULL, uts? &t :NULL, sigsetsize);
 	set_fs(old_fs);
 	if (ret >= 0 && uinfo) {
 		if (copy_siginfo_to_user32(uinfo, &info))

Re: ia32 sigwait patch

[Andi Kleen]

On Wed, 3 Mar 2004 16:06:16 -0700
Bjorn Helgaas <bjorn.helgaas@hp.com> wrote:

> It looks to me like current 2.4 on x86_64 suffers from the same
> problem with uts (sys_rt_sigtimedwait() gets a pointer to
> uninitialized "t" when uts is NULL).

Yes, indeed. Thanks for the hint.

-Andi