[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20090928045131.GA15149@localhost>
Date: Mon, 28 Sep 2009 12:51:31 +0800
From: Wu Fengguang <fengguang.wu@...el.com>
To: Nigel Cunningham <ncunningham@...a.org.au>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>,
LKML <linux-kernel@...r.kernel.org>,
"linux-mm@...ck.org" <linux-mm@...ck.org>
Subject: Re: No more bits in vm_area_struct's vm_flags.
On Mon, Sep 28, 2009 at 12:37:48PM +0800, Nigel Cunningham wrote:
> Hi.
>
> KAMEZAWA Hiroyuki wrote:
> > Then, Nigel, you have 2 choices I think.
> >
> > (1) don't merge if vm_hints is set or (2) pass vm_hints to all
> > __merge() functions.
> >
> > One of above will be accesptable for stakeholders... I personally
> > like (1) but just trying (2) may be accepted.
> >
> > What I dislike is making vm_flags to be long long ;)
>
> Okay. I've gone for option 1 for now. Here's what I
> currently have (compile testing as I write)...
>
>
>
> vm_flags in struct vm_area_struct is full. Move some of the less commonly
> used flags to a new variable so that other flags that need to be in vm_flags
> (because, for example, they need to be in variables that are passed around)
> can be added.
Looks good to me with some minor suggestions.
Reviewed-by: Wu Fengguang <fengguang.wu@...el.com>
> Signed-off-by: Nigel Cunningham <nigel@...onice.net>
> ---
> include/linux/mm.h | 16 ++++++++--------
> include/linux/mm_types.h | 1 +
> mm/madvise.c | 28 ++++++++++++++++++----------
> mm/mmap.c | 2 ++
> 4 files changed, 29 insertions(+), 18 deletions(-)
>
> diff --git a/include/linux/mm.h b/include/linux/mm.h
> index 24c3956..040d0ce 100644
> --- a/include/linux/mm.h
> +++ b/include/linux/mm.h
> @@ -85,10 +85,6 @@ extern unsigned int kobjsize(const void *objp);
> #define VM_LOCKED 0x00002000
> #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
>
> - /* Used by sys_madvise() */
> -#define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
> -#define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
> -
> #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
> #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
> #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
> @@ -116,11 +112,15 @@ extern unsigned int kobjsize(const void *objp);
> #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
> #endif
Maybe add this comment for less confusion?
/*
* vm_hints in vm_area_struct, see mm_types.h.
*/
> + /* Used by sys_madvise() */
> +#define VM_SEQ_READ 0x00000001 /* App will access data sequentially */
> +#define VM_RAND_READ 0x00000002 /* App will not benefit from clustered reads */
> +
> #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
> -#define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
> -#define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
> -#define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
> -#define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
> +#define VM_ClearReadHint(v) (v)->vm_hints &= ~VM_READHINTMASK
> +#define VM_NormalReadHint(v) (!((v)->vm_hints & VM_READHINTMASK))
> +#define VM_SequentialReadHint(v) ((v)->vm_hints & VM_SEQ_READ)
> +#define VM_RandomReadHint(v) ((v)->vm_hints & VM_RAND_READ)
>
> /*
> * special vmas that are non-mergable, non-mlock()able
> diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
> index 84a524a..5c66e3a 100644
> --- a/include/linux/mm_types.h
> +++ b/include/linux/mm_types.h
> @@ -178,6 +178,7 @@ struct vm_area_struct {
> units, *not* PAGE_CACHE_SIZE */
> struct file * vm_file; /* File we map to (can be NULL). */
> void * vm_private_data; /* was vm_pte (shared mem) */
> + unsigned long vm_hints; /* Hints, see mm.h. */
> unsigned long vm_truncate_count;/* truncate_count or restart_addr */
>
> #ifndef CONFIG_MMU
> diff --git a/mm/madvise.c b/mm/madvise.c
> index 35b1479..59a93d3 100644
> --- a/mm/madvise.c
> +++ b/mm/madvise.c
> @@ -40,19 +40,22 @@ static long madvise_behavior(struct vm_area_struct * vma,
> unsigned long start, unsigned long end, int behavior)
> {
> struct mm_struct * mm = vma->vm_mm;
> - int error = 0;
> + int error = 0, skip_merge = 0;
> pgoff_t pgoff;
> - unsigned long new_flags = vma->vm_flags;
> + unsigned long new_flags = vma->vm_flags, new_hints = vma->vm_hints;
It would be nicer to add standalone lines for skip_merge and new_hints.
Thanks,
Fengguang
>
> switch (behavior) {
> case MADV_NORMAL:
> - new_flags = new_flags & ~VM_RAND_READ & ~VM_SEQ_READ;
> + new_hints = new_hints & ~VM_RAND_READ & ~VM_SEQ_READ;
> + skip_merge = 1;
> break;
> case MADV_SEQUENTIAL:
> - new_flags = (new_flags & ~VM_RAND_READ) | VM_SEQ_READ;
> + new_hints = (new_hints & ~VM_RAND_READ) | VM_SEQ_READ;
> + skip_merge = 1;
> break;
> case MADV_RANDOM:
> - new_flags = (new_flags & ~VM_SEQ_READ) | VM_RAND_READ;
> + new_hints = (new_hints & ~VM_SEQ_READ) | VM_RAND_READ;
> + skip_merge = 1;
> break;
> case MADV_DONTFORK:
> new_flags |= VM_DONTCOPY;
> @@ -78,11 +81,15 @@ static long madvise_behavior(struct vm_area_struct * vma,
> }
>
> pgoff = vma->vm_pgoff + ((start - vma->vm_start) >> PAGE_SHIFT);
> - *prev = vma_merge(mm, *prev, start, end, new_flags, vma->anon_vma,
> - vma->vm_file, pgoff, vma_policy(vma));
> - if (*prev) {
> - vma = *prev;
> - goto success;
> +
> + if (!skip_merge) {
> + *prev = vma_merge(mm, *prev, start, end, new_flags,
> + vma->anon_vma, vma->vm_file, pgoff,
> + vma_policy(vma));
> + if (*prev) {
> + vma = *prev;
> + goto success;
> + }
> }
>
> *prev = vma;
> @@ -104,6 +111,7 @@ success:
> * vm_flags is protected by the mmap_sem held in write mode.
> */
> vma->vm_flags = new_flags;
> + vma->vm_hints = new_hints;
>
> out:
> if (error == -ENOMEM)
> diff --git a/mm/mmap.c b/mm/mmap.c
> index 73f5e4b..fb4bf98 100644
> --- a/mm/mmap.c
> +++ b/mm/mmap.c
> @@ -670,6 +670,8 @@ static inline int is_mergeable_vma(struct vm_area_struct *vma,
> return 0;
> if (vma->vm_ops && vma->vm_ops->close)
> return 0;
> + if (vma->vm_hints)
> + return 0;
> return 1;
> }
>
> --
> 1.6.3.3
--
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/
Powered by blists - more mailing lists