linux-kernel - Re: [PATCH] [6/8] Account overlapped mappings in end_pfn

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20080211130843.GC23733@elte.hu>
Date:	Mon, 11 Feb 2008 14:08:43 +0100
From:	Ingo Molnar <mingo@...e.hu>
To:	Andi Kleen <ak@...e.de>
Cc:	ying.huang@...el.com, tglx@...utronix.de,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH] [6/8] Account overlapped mappings in end_pfn_map

* Andi Kleen <ak@...e.de> wrote:

> When end_pfn is not aligned to 2MB (or 1GB) then the kernel might map 
> more memory than end_pfn. Account this in end_pfn_mapped.

can you see any practical relevance? Your patch description only deals 
with the mechanical details of the change instead of analyzing the most 
important thing: relevance and impact analysis. That makes it harder to 
add your patches and easier to miss a good fix accidentally. It also 
makes it quite a bit harder to trust your patches.

at a quick glance the relevance is to EFI only, in 
efi_enter_virtual_mode(). max_pfn_mapped is used as a differentiator 
between __va() and efi_ioremap(). AFAICS EFI will typically have its 
runtime code not right after the end of physical memory.

Nevertheless, i do agree that the max_pfn_mapped/end_pfn_map limit needs 
to be sharpened to reflect reality (for later PAT support).

your patch is also a bit unclean:

> -static void __meminit
> +static unsigned long __meminit
>  phys_pmd_update(pud_t *pud, unsigned long address, unsigned long end)
>  {
> +	unsigned long r;
>  	pmd_t *pmd = pmd_offset(pud, 0);
>  	spin_lock(&init_mm.page_table_lock);
> -	phys_pmd_init(pmd, address, end);
> +	r = phys_pmd_init(pmd, address, end);
>  	spin_unlock(&init_mm.page_table_lock);
>  	__flush_tlb_all();
> +	return r;
>  }

a meaningless "r" parameter. Use "true_end" instead.

and here's another type of variable naming problem:

> +unsigned long __init_refok
> +init_memory_mapping(unsigned long start, unsigned long end)
>  {
> -	unsigned long next;
> +	unsigned long next, true_end = end;

that "true_end" is already a _pfn_, so name it accordingly: 
true_end_pfn. Because we switch between pfn and addr types it's 
important to point out where it's pfn and where it's a linear/physical 
address.

>  	struct zone *zone = pgdat->node_zones + ZONE_NORMAL;
>  	unsigned long start_pfn = start >> PAGE_SHIFT;
>  	unsigned long nr_pages = size >> PAGE_SHIFT;
> +	unsigned long r;
>  	int ret;
>  
> -	init_memory_mapping(start, start + size-1);
> +	r = init_memory_mapping(start, start + size-1);
> +	if (r > end_pfn_map)
> +		end_pfn_map = r;

rename "r" to "true_end_pfn".

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