[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-arch
Subject:    Re: [PATCH 1/5] arm64: compat: Alloc separate pages for vectors and sigpage
From:       Catalin Marinas <catalin.marinas () arm ! com>
Date:       2019-03-29 12:12:14
Message-ID: 20190329121211.gv7i7fqt74h3rtls () mbp
[Download RAW message or body]

On Tue, Mar 19, 2019 at 03:15:38PM +0000, Vincenzo Frascino wrote:
> +static int __init aarch32_alloc_vdso_pages(void)
>  {
>  	extern char __kuser_helper_start[], __kuser_helper_end[];
>  	extern char __aarch32_sigret_code_start[], __aarch32_sigret_code_end[];
>  
>  	int kuser_sz = __kuser_helper_end - __kuser_helper_start;
>  	int sigret_sz = __aarch32_sigret_code_end - __aarch32_sigret_code_start;
> -	unsigned long vpage;
> +	unsigned long vdso_pages[2];
>  
> -	vpage = get_zeroed_page(GFP_ATOMIC);
> +	vdso_pages[0] = get_zeroed_page(GFP_ATOMIC);
> +	if (!vdso_pages[0])
> +		return -ENOMEM;
>  
> -	if (!vpage)
> +	vdso_pages[1] = get_zeroed_page(GFP_ATOMIC);
> +	if (!vdso_pages[1])
>  		return -ENOMEM;
>  
>  	/* kuser helpers */
> -	memcpy((void *)vpage + 0x1000 - kuser_sz, __kuser_helper_start,
> -		kuser_sz);
> +	memcpy((void *)(vdso_pages[0] + 0x1000 - kuser_sz),
> +	       __kuser_helper_start,
> +	       kuser_sz);
>  
>  	/* sigreturn code */
> -	memcpy((void *)vpage + AARCH32_KERN_SIGRET_CODE_OFFSET,
> -               __aarch32_sigret_code_start, sigret_sz);
> +	memcpy((void *)vdso_pages[1],
> +	       __aarch32_sigret_code_start,
> +	       sigret_sz);
>  
> -	flush_icache_range(vpage, vpage + PAGE_SIZE);
> -	vectors_page[0] = virt_to_page(vpage);
> +	flush_icache_range(vdso_pages[0], vdso_pages[0] + PAGE_SIZE);
> +	flush_icache_range(vdso_pages[1], vdso_pages[1] + PAGE_SIZE);

I wonder whether we still need the flush_icache_range() calls here.
Normally set_pte_at() handles cache maintenance automatically if
PG_dcache_clean is not set. We could probably just call
flush_dcache_page() here.

-- 
Catalin
[prev in list] [next in list] [prev in thread] [next in thread]