Re: [PATCH 2.6.26-rc4] Fix nesting error checking in arch/x86/mm/ioremap.c

[Ingo Molnar <mingo@elte.hu>]

* Mika Kukkonen <mikukkon@srv1-m700-lanp.koti> wrote:

>  	unsigned int nesting;
>  
> -	nesting = --early_ioremap_nested;
> -	WARN_ON(nesting < 0);
> +	if (early_ioremap_nested)
> +		nesting = --early_ioremap_nested;
> +	else { 
> +		nesting = 0;
> +		WARN_ON(1);
> +	}

the better fix would be to turn 'nesting' into an int, hm?

Also, instead of forcing nesting to 0, it is better to just return - as 
we cannot do any sane unmap here anymore. (we are unbalanced)

I.e. something like the fix below?

	Ingo

------------------------------>
Subject: ioremap: nesting fix
From: Ingo Molnar <mingo@elte.hu>
Date: Tue May 27 09:56:49 CEST 2008

Mika Kukkonen noticed that the nesting check in early_iounmap() is not 
actually done.

Reported-by: Mika Kukkonen <mikukkon@srv1-m700-lanp.koti>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
---
 arch/x86/mm/ioremap.c |    5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

Index: linux/arch/x86/mm/ioremap.c
===================================================================
--- linux.orig/arch/x86/mm/ioremap.c
+++ linux/arch/x86/mm/ioremap.c
@@ -602,10 +602,11 @@ void __init early_iounmap(void *addr, un
 	unsigned long offset;
 	unsigned int nrpages;
 	enum fixed_addresses idx;
-	unsigned int nesting;
+	int nesting;
 
 	nesting = --early_ioremap_nested;
-	WARN_ON(nesting < 0);
+	if (WARN_ON(nesting < 0))
+		return;
 
 	if (early_ioremap_debug) {
 		printk(KERN_INFO "early_iounmap(%p, %08lx) [%d]\n", addr,