Re: 2.6.14 kernels and above copy_to_user stupidity with IRQ disabledcheck

From: Jeff V. Merkey
Date: Fri Jan 27 2006 - 16:17:04 EST


Phillip Susi wrote:

jmerkey@xxxxxxxxxxxxxxxx wrote:

OK. Got it. I guess I need to restructure. And BTW, This was a code fragment
only, the spinlock gets released when -EFAULT is called -- was just an example.

Jeff


Unless you have redefined EFAULT in some strange and hideous way, it is not "called" and doesn't free the spinlock. EFAULT is defined as a literal integer, so you're just returning a number without freeing the spinlock.

If you have redefined EFAULT to a macro function call or whatever, then don't do that, it's REALLY horrible coding practice.


No. I posted a code fragment as an example. Here's the actual code:

int dump_regen(VIRTUAL_SETUP *s, ULONG count)
{
register int i = 0;
VIRTUAL_SETUP *v;


spin_lock_irqsave(&regen_lock, regen_flags);
v = regen_head;
while (v)
{
if (i >= count)
{
spin_unlock_irqrestore(&regen_lock, regen_flags);
return -EFAULT;
}


err = copy_to_user(&s[i++], v, sizeof(VIRTUAL_SETUP));
if (err)
{ spin_unlock_irqrestore(&regen_lock, regen_flags);
return err;
}

v = v->next;
}
spin_unlock_irqrestore(&regen_lock, regen_flags);
return 0;
}

Needless to say, this has been restructured to this:

int dump_regen(VIRTUAL_SETUP *s, ULONG count)
{
register int i = 0;
VIRTUAL_SETUP *v;


spin_lock_irqsave(&regen_lock, regen_flags);
v = regen_head;
while (v)
{
if (i >= count)
{
spin_unlock_irqrestore(&regen_lock, regen_flags);
return 0;
}


P_Copy(&s[i++], v, sizeof(VIRTUAL_SETUP));
v = v->next;
}
spin_unlock_irqrestore(&regen_lock, regen_flags);
return 0;
}


Jeff

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/