linux-kernel - Re: [PATCH] mm: extend zero pages to same element pages for zram

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: <20170206234805.GA12188@bbox>
Date:   Tue, 7 Feb 2017 08:48:05 +0900
From:   Minchan Kim <minchan@...nel.org>
To:     zhouxianrong <zhouxianrong@...wei.com>
CC:     <linux-mm@...ck.org>, <linux-kernel@...r.kernel.org>,
        <akpm@...ux-foundation.org>, <sergey.senozhatsky@...il.com>,
        <willy@...radead.org>, <iamjoonsoo.kim@....com>,
        <ngupta@...are.org>, <Mi.Sophia.Wang@...wei.com>,
        <zhouxiyu@...wei.com>, <weidu.du@...wei.com>,
        <zhangshiming5@...wei.com>, <won.ho.park@...wei.com>
Subject: Re: [PATCH] mm: extend zero pages to same element pages for zram

Hi

On Mon, Feb 06, 2017 at 09:28:18AM +0800, zhouxianrong wrote:
> 
> 
> On 2017/2/5 22:21, Minchan Kim wrote:
> >Hi zhouxianrong,
> >
> >On Fri, Feb 03, 2017 at 04:42:27PM +0800, zhouxianrong@...wei.com wrote:
> >>From: zhouxianrong <zhouxianrong@...wei.com>
> >>
> >>test result as listed below:
> >>
> >>zero   pattern_char pattern_short pattern_int pattern_long total   (unit)
> >>162989 14454        3534          23516       2769         3294399 (page)
> >>
> >>statistics for the result:
> >>
> >>        zero  pattern_char  pattern_short  pattern_int  pattern_long
> >>AVERAGE 0.745696298 0.085937175 0.015957701 0.131874915 0.020533911
> >>STDEV   0.035623777 0.016892402 0.004454534 0.021657123 0.019420072
> >>MAX     0.973813421 0.222222222 0.021409518 0.211812245 0.176512625
> >>MIN     0.645431905 0.004634398 0           0           0
> >
> >The description in old version was better for justifying same page merging
> >feature.
> >
> >>
> >>Signed-off-by: zhouxianrong <zhouxianrong@...wei.com>
> >>---
> >> drivers/block/zram/zram_drv.c |  124 +++++++++++++++++++++++++++++++----------
> >> drivers/block/zram/zram_drv.h |   11 ++--
> >> 2 files changed, 103 insertions(+), 32 deletions(-)
> >>
> >>diff --git a/drivers/block/zram/zram_drv.c b/drivers/block/zram/zram_drv.c
> >>index e5ab7d9..6a8c9c5 100644
> >>--- a/drivers/block/zram/zram_drv.c
> >>+++ b/drivers/block/zram/zram_drv.c
> >>@@ -95,6 +95,17 @@ static void zram_clear_flag(struct zram_meta *meta, u32 index,
> >> 	meta->table[index].value &= ~BIT(flag);
> >> }
> >>
> >>+static inline void zram_set_element(struct zram_meta *meta, u32 index,
> >>+			unsigned long element)
> >>+{
> >>+	meta->table[index].element = element;
> >>+}
> >>+
> >>+static inline void zram_clear_element(struct zram_meta *meta, u32 index)
> >>+{
> >>+	meta->table[index].element = 0;
> >>+}
> >>+
> >> static size_t zram_get_obj_size(struct zram_meta *meta, u32 index)
> >> {
> >> 	return meta->table[index].value & (BIT(ZRAM_FLAG_SHIFT) - 1);
> >>@@ -167,31 +178,78 @@ static inline void update_used_max(struct zram *zram,
> >> 	} while (old_max != cur_max);
> >> }
> >>
> >>-static bool page_zero_filled(void *ptr)
> >>+static inline void zram_fill_page(char *ptr, unsigned long value)
> >>+{
> >>+	int i;
> >>+	unsigned long *page = (unsigned long *)ptr;
> >>+
> >>+	if (likely(value == 0)) {
> >>+		clear_page(ptr);
> >>+	} else {
> >>+		for (i = 0; i < PAGE_SIZE / sizeof(*page); i++)
> >>+			page[i] = value;
> >>+	}
> >>+}
> >>+
> >>+static inline void zram_fill_page_partial(char *ptr, unsigned int size,
> >>+		unsigned long value)
> >>+{
> >>+	int i;
> >>+	unsigned long *page;
> >>+
> >>+	if (likely(value == 0)) {
> >>+		memset(ptr, 0, size);
> >>+		return;
> >>+	}
> >>+
> >>+	i = ((unsigned long)ptr) % sizeof(*page);
> >>+	if (i) {
> >>+		while (i < sizeof(*page)) {
> >>+			*ptr++ = (value >> (i * 8)) & 0xff;
> >>+			--size;
> >>+			++i;
> >>+		}
> >>+	}
> >>+
> >
> >I don't think we need this part because block layer works with sector
> >size or multiple times of it so it must be aligned unsigned long.
> >
> >
> >
> >
> >.
> >
> 
> Minchan and Matthew Wilcox:
> 
> 1. right, but users could open /dev/block/zram0 file and do any read operations.

Could you make that happen?
I don't think it's possible as Matthew already pointed out, too.

> 
> 2. about endian operation for long, the modification is trivial and low efficient.
>    i have not better method. do you have any good idea for this?

So if assumption 1 is wrong, we don't need 2, either.

> 
> 3. the below should be modified.
> 
> static inline bool zram_meta_get(struct zram *zram)
> @@ -495,11 +553,17 @@ static void zram_meta_free(struct zram_meta *meta, u64 disksize)
> 
>  	/* Free all pages that are still in this zram device */
>  	for (index = 0; index < num_pages; index++) {
> -		unsigned long handle = meta->table[index].handle;
> +		unsigned long handle;
> +
> +		bit_spin_lock(ZRAM_ACCESS, &meta->table[index].value);
> +		handle = meta->table[index].handle;
> 
> -		if (!handle)
> +		if (!handle || zram_test_flag(meta, index, ZRAM_SAME)) {
> +			bit_spin_unlock(ZRAM_ACCESS, &meta->table[index].value);
>  			continue;
> +		}
> 
> +		bit_spin_unlock(ZRAM_ACCESS, &meta->table[index].value);
>  		zs_free(meta->mem_pool, handle);

Could you explain why we need this modification?

>  	}
> 
> @@ -511,7 +575,7 @@ static void zram_meta_free(struct zram_meta *meta, u64 disksize)
>  static struct zram_meta *zram_meta_alloc(char *pool_name, u64 disksize)
>  {
>  	size_t num_pages;
> -	struct zram_meta *meta = kmalloc(sizeof(*meta), GFP_KERNEL);
> +	struct zram_meta *meta = kzalloc(sizeof(*meta), GFP_KERNEL);

Ditto

> 
>