linux-kernel - Re: [PATCH] LoongArch: Remove shm_align

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: <08503cb1-102e-9101-51a1-47b7cf7cb0be@xen0n.name>
Date:   Tue, 29 Aug 2023 01:05:20 +0800
From:   WANG Xuerui <kernel@...0n.name>
To:     Huacai Chen <chenhuacai@...ngson.cn>,
        Arnd Bergmann <arnd@...db.de>,
        Huacai Chen <chenhuacai@...nel.org>
Cc:     loongarch@...ts.linux.dev, linux-arch@...r.kernel.org,
        Xuefeng Li <lixuefeng@...ngson.cn>,
        Guo Ren <guoren@...nel.org>,
        Jiaxun Yang <jiaxun.yang@...goat.com>,
        linux-kernel@...r.kernel.org, loongson-kernel@...ts.loongnix.cn,
        Jiantao Shan <shanjiantao@...ngson.cn>
Subject: Re: [PATCH] LoongArch: Remove shm_align_mask and use SHMLBA instead

On 8/28/23 23:25, Huacai Chen wrote:
> Both shm_align_mask and SHMLBA want to avoid cache alias. But they are
> inconsistent: shm_align_mask is (PAGE_SIZE - 1) while SHMLBA is SZ_64K,
> but PAGE_SIZE is not always equal to SZ_64K.
>
> This may cause problems when shmat() twice. Fix this problem by removing
> shm_align_mask and using SHMLBA (SHMLBA - 1, strctly) instead.
"strictly SHMLBA - 1"?
>
> Reported-by: Jiantao Shan <shanjiantao@...ngson.cn>
> Signed-off-by: Huacai Chen <chenhuacai@...ngson.cn>
> ---
>   arch/loongarch/mm/cache.c      |  1 -
>   arch/loongarch/mm/mmap.c       | 12 ++++--------
>   3 files changed, 8 insertions(+), 9 deletions(-)
>
> diff --git a/arch/loongarch/mm/cache.c b/arch/loongarch/mm/cache.c
> index 72685a48eaf0..6be04d36ca07 100644
> --- a/arch/loongarch/mm/cache.c
> +++ b/arch/loongarch/mm/cache.c
> @@ -156,7 +156,6 @@ void cpu_cache_init(void)
>   
>   	current_cpu_data.cache_leaves_present = leaf;
>   	current_cpu_data.options |= LOONGARCH_CPU_PREFETCH;
> -	shm_align_mask = PAGE_SIZE - 1;
>   }
>   
>   static const pgprot_t protection_map[16] = {
> diff --git a/arch/loongarch/mm/mmap.c b/arch/loongarch/mm/mmap.c
> index fbe1a4856fc4..c99c8015651a 100644
> --- a/arch/loongarch/mm/mmap.c
> +++ b/arch/loongarch/mm/mmap.c
> @@ -8,12 +8,8 @@
>   #include <linux/mm.h>
>   #include <linux/mman.h>
>   
> -unsigned long shm_align_mask = PAGE_SIZE - 1;	/* Sane caches */
> -EXPORT_SYMBOL(shm_align_mask);

By removing this altogether, a lot of code duplication is introduced it 
seems. Better make this a private #define so use sites remain nicely 
symbolic:

"#define SHM_ALIGN_MASK (SHMLBA - 1)"

> -
> -#define COLOUR_ALIGN(addr, pgoff)				\
> -	((((addr) + shm_align_mask) & ~shm_align_mask) +	\
> -	 (((pgoff) << PAGE_SHIFT) & shm_align_mask))
> +#define COLOUR_ALIGN(addr, pgoff)			\
> +	((((addr) + (SHMLBA - 1)) & ~(SHMLBA - 1)) + (((pgoff) << PAGE_SHIFT) & (SHMLBA - 1)))
>   
>   enum mmap_allocation_direction {UP, DOWN};
>   
> @@ -40,7 +36,7 @@ static unsigned long arch_get_unmapped_area_common(struct file *filp,
>   		 * cache aliasing constraints.
>   		 */
>   		if ((flags & MAP_SHARED) &&
> -		    ((addr - (pgoff << PAGE_SHIFT)) & shm_align_mask))
> +		    ((addr - (pgoff << PAGE_SHIFT)) & (SHMLBA - 1)))
>   			return -EINVAL;
>   		return addr;
>   	}
> @@ -63,7 +59,7 @@ static unsigned long arch_get_unmapped_area_common(struct file *filp,
>   	}
>   
>   	info.length = len;
> -	info.align_mask = do_color_align ? (PAGE_MASK & shm_align_mask) : 0;
> +	info.align_mask = do_color_align ? (PAGE_MASK & (SHMLBA - 1)) : 0;
>   	info.align_offset = pgoff << PAGE_SHIFT;
>   
>   	if (dir == DOWN) {

-- 
WANG "xen0n" Xuerui

Linux/LoongArch mailing list: https://lore.kernel.org/loongarch/