No more bits in vm_area_struct's vm_flags.

No more bits in vm_area_struct's vm_flags.

[Nigel Cunningham]

Hi all.

With the addition of the VM_MERGEABLE flag to vm_flags post-2.6.31, the
last bit in vm_flags has been used.

I have some code in TuxOnIce that needs a bit too (explicitly mark the
VMA as needing to be atomically copied, for GEM objects), and am not
sure what the canonical way to proceed is. Should a new unsigned long be
added? The difficulty I see with that is that my flag was used in
shmem_file_setup's flags parameter (drm_gem_object_alloc), so that
function would need an extra parameter too..

Regards,

Nigel

Re: No more bits in vm_area_struct's vm_flags.

[Hugh Dickins]

On Wed, 23 Sep 2009, Nigel Cunningham wrote:
> 
> With the addition of the VM_MERGEABLE flag to vm_flags post-2.6.31, the
> last bit in vm_flags has been used.

Yes, it was rather selfish to take that, without even pointing out
that was the last of 32 bits in the changelog, and without mapping
out where to go next - sorry.

> 
> I have some code in TuxOnIce that needs a bit too (explicitly mark the
> VMA as needing to be atomically copied, for GEM objects), and am not

(I wonder what atomically copied means there.)

> sure what the canonical way to proceed is. Should a new unsigned long be
> added? The difficulty I see with that is that my flag was used in
> shmem_file_setup's flags parameter (drm_gem_object_alloc), so that
> function would need an extra parameter too..

I've assumed that, when necessary, we'll retype vm_flags from
unsigned long to unsigned long long (or u64).  But I've not yet
checked how much bloat that would add to 32-bit code: whether we
should put it off as long as we can, or be lazy and do it soon.

I'm thinking that we should use the full 32-bit vm_flags as a
prompt to dispose of a few.  VM_RESERVED is the one I always claim
I'm going to remove, then more important jobs intervene; and we seem
to have grown more weird variants of VM_PFNMAP than I care for in
the last year or two.  Suspect VM_PFN_AT_MMAP could make reasonable
use of VM_NONLINEAR, but probably not without some small change.

Does TuxOnIce rely on CONFIG_MMU?  If so, then the TuxOnIce patch
could presumably reuse VM_MAPPED_COPY for now - but don't be
surprised if that's one we clean away later on.

Hugh

Re: No more bits in vm_area_struct's vm_flags.

[KAMEZAWA Hiroyuki]

On Wed, 23 Sep 2009 14:15:18 +1000
Nigel Cunningham <ncunningham@crca.org.au> wrote:

> Hi all.
> 
> With the addition of the VM_MERGEABLE flag to vm_flags post-2.6.31, the
> last bit in vm_flags has been used.
> 

Wow...that's bad.



> I have some code in TuxOnIce that needs a bit too (explicitly mark the
> VMA as needing to be atomically copied, for GEM objects), and am not
> sure what the canonical way to proceed is. Should a new unsigned long be
> added? The difficulty I see with that is that my flag was used in
> shmem_file_setup's flags parameter (drm_gem_object_alloc), so that
> function would need an extra parameter too..
> 

Hmm, how about adding vma->vm_flags2 ?

Thanks,
-Kame


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

Re: No more bits in vm_area_struct's vm_flags.

[Nigel Cunningham]

Hi.

Hugh Dickins wrote:
> On Wed, 23 Sep 2009, Nigel Cunningham wrote:
>> With the addition of the VM_MERGEABLE flag to vm_flags post-2.6.31, the
>> last bit in vm_flags has been used.
> 
> Yes, it was rather selfish to take that, without even pointing out
> that was the last of 32 bits in the changelog, and without mapping
> out where to go next - sorry.

Not at all! I'm only asking because I wasn't sure where to go next.

>> I have some code in TuxOnIce that needs a bit too (explicitly mark the
>> VMA as needing to be atomically copied, for GEM objects), and am not
> 
> (I wonder what atomically copied means there.)

Copied together, with interrupts disabled (to get as consistent an image
as possible). Swsusp and uswsusp atomically copy everything, and so
don't need this flag. TuxOnIce relies on the freezer to stop userspace
pages changing, and saves them separately in order to get a full image
of memory. The flag is needed because these pages would otherwise not be
included in the set of pages that are atomic copied.

>> sure what the canonical way to proceed is. Should a new unsigned long be
>> added? The difficulty I see with that is that my flag was used in
>> shmem_file_setup's flags parameter (drm_gem_object_alloc), so that
>> function would need an extra parameter too..
> 
> I've assumed that, when necessary, we'll retype vm_flags from
> unsigned long to unsigned long long (or u64).  But I've not yet
> checked how much bloat that would add to 32-bit code: whether we
> should put it off as long as we can, or be lazy and do it soon.
> 
> I'm thinking that we should use the full 32-bit vm_flags as a
> prompt to dispose of a few.  VM_RESERVED is the one I always claim
> I'm going to remove, then more important jobs intervene; and we seem
> to have grown more weird variants of VM_PFNMAP than I care for in
> the last year or two.  Suspect VM_PFN_AT_MMAP could make reasonable
> use of VM_NONLINEAR, but probably not without some small change.
> 
> Does TuxOnIce rely on CONFIG_MMU?  If so, then the TuxOnIce patch
> could presumably reuse VM_MAPPED_COPY for now - but don't be
> surprised if that's one we clean away later on.

Hmm. I'm not sure. The requirements are the same as for swsusp and
uswsusp. Is there some tool to graph config dependencies?

Regards,

Nigel

Re: No more bits in vm_area_struct's vm_flags.

[Nigel Cunningham]

Hi.

KAMEZAWA Hiroyuki wrote:
>> I have some code in TuxOnIce that needs a bit too (explicitly mark the
>> VMA as needing to be atomically copied, for GEM objects), and am not
>> sure what the canonical way to proceed is. Should a new unsigned long be
>> added? The difficulty I see with that is that my flag was used in
>> shmem_file_setup's flags parameter (drm_gem_object_alloc), so that
>> function would need an extra parameter too..
> 
> Hmm, how about adding vma->vm_flags2 ?

The difficulty there is that some functions pass these flags as arguments.

Regards,

Nigel

Re: No more bits in vm_area_struct's vm_flags.

[KAMEZAWA Hiroyuki]

On Fri, 25 Sep 2009 18:34:56 +1000
Nigel Cunningham <ncunningham@crca.org.au> wrote:

> Hi.
> 
> KAMEZAWA Hiroyuki wrote:
> >> I have some code in TuxOnIce that needs a bit too (explicitly mark the
> >> VMA as needing to be atomically copied, for GEM objects), and am not
> >> sure what the canonical way to proceed is. Should a new unsigned long be
> >> added? The difficulty I see with that is that my flag was used in
> >> shmem_file_setup's flags parameter (drm_gem_object_alloc), so that
> >> function would need an extra parameter too..
> > 
> > Hmm, how about adding vma->vm_flags2 ?
> 
> The difficulty there is that some functions pass these flags as arguments.
> 
Ah yes. But I wonder some special flags, which is rarey used, can be moved
to vm_flags2...

For example,

 #define VM_SEQ_READ     0x00008000      /* App will access data sequentially */
 #define VM_RAND_READ    0x00010000      /* App will not benefit from clustered reads */
are all capsuled under
mm.h
 117 #define VM_READHINTMASK                 (VM_SEQ_READ | VM_RAND_READ)
 118 #define VM_ClearReadHint(v)             (v)->vm_flags &= ~VM_READHINTMASK
 119 #define VM_NormalReadHint(v)            (!((v)->vm_flags & VM_READHINTMASK))
 120 #define VM_SequentialReadHint(v)        ((v)->vm_flags & VM_SEQ_READ)
 121 #define VM_RandomReadHint(v)            ((v)->vm_flags & VM_RAND_READ)

Or

105 #define VM_PFN_AT_MMAP  0x40000000      /* PFNMAP vma that is fully mapped at mmap time */
is only used under special situation.

etc..

They'll be able to be moved to other(new) flag field, IIUC.

Thanks,
-Kame




> Regards,
> 
> Nigel
> 
> --
> To unsubscribe, send a message with 'unsubscribe linux-mm' in
> the body to majordomo@kvack.org.  For more info on Linux MM,
> see: http://www.linux-mm.org/ .
> Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
> 

Re: No more bits in vm_area_struct's vm_flags.

[Nigel Cunningham]

Hi.

KAMEZAWA Hiroyuki wrote:
> On Fri, 25 Sep 2009 18:34:56 +1000
> Nigel Cunningham <ncunningham@crca.org.au> wrote:
> 
>> Hi.
>>
>> KAMEZAWA Hiroyuki wrote:
>>>> I have some code in TuxOnIce that needs a bit too (explicitly mark the
>>>> VMA as needing to be atomically copied, for GEM objects), and am not
>>>> sure what the canonical way to proceed is. Should a new unsigned long be
>>>> added? The difficulty I see with that is that my flag was used in
>>>> shmem_file_setup's flags parameter (drm_gem_object_alloc), so that
>>>> function would need an extra parameter too..
>>> Hmm, how about adding vma->vm_flags2 ?
>> The difficulty there is that some functions pass these flags as arguments.
>>
> Ah yes. But I wonder some special flags, which is rarey used, can be moved
> to vm_flags2...

Ah, of course. That makes sense.

Regards,

Nigel

Re: No more bits in vm_area_struct's vm_flags.

[Hugh Dickins]

On Fri, 25 Sep 2009, Nigel Cunningham wrote:
> Hugh Dickins wrote:
> > 
> > Does TuxOnIce rely on CONFIG_MMU?  If so, then the TuxOnIce patch
> > could presumably reuse VM_MAPPED_COPY for now - but don't be
> > surprised if that's one we clean away later on.
> 
> Hmm. I'm not sure. The requirements are the same as for swsusp and
> uswsusp. Is there some tool to graph config dependencies?

If its requirements are the same as theirs, then I think yes,
TuxOnIce relies on  config MMU: because config HIBERNATION depends
on SWAP (kernel/power/Kconfig) and config SWAP depends on MMU
(init/Kconfig).

But I hesitated to assume that, because in the TuxOnIce patch I have
here for reference (e.g. when deciding it'll cause you unwelcome hassle
if I move swap_info_struct from swap.h to swapfile.c!), it looked as if
you support saving to an ordinary file, not just to a swap file?

> Is there some tool to graph config dependencies?

I don't know it, but would be nice, and maybe it does exist.

Hugh

Re: No more bits in vm_area_struct's vm_flags.

[Nigel Cunningham]

Hi.

Hugh Dickins wrote:
> On Fri, 25 Sep 2009, Nigel Cunningham wrote:
>> Hugh Dickins wrote:
>>> Does TuxOnIce rely on CONFIG_MMU?  If so, then the TuxOnIce patch
>>> could presumably reuse VM_MAPPED_COPY for now - but don't be
>>> surprised if that's one we clean away later on.
>> Hmm. I'm not sure. The requirements are the same as for swsusp and
>> uswsusp. Is there some tool to graph config dependencies?
> 
> If its requirements are the same as theirs, then I think yes,
> TuxOnIce relies on  config MMU: because config HIBERNATION depends
> on SWAP (kernel/power/Kconfig) and config SWAP depends on MMU
> (init/Kconfig).
> 
> But I hesitated to assume that, because in the TuxOnIce patch I have
> here for reference (e.g. when deciding it'll cause you unwelcome hassle
> if I move swap_info_struct from swap.h to swapfile.c!), it looked as if
> you support saving to an ordinary file, not just to a swap file?

Yes, that's right.

>> Is there some tool to graph config dependencies?
> 
> I don't know it, but would be nice, and maybe it does exist.

Okee doke.

Nigel

Re: No more bits in vm_area_struct's vm_flags.

[Joerg Roedel]

On Fri, Sep 25, 2009 at 06:48:34PM +1000, Nigel Cunningham wrote:
> Hi.
> 
> KAMEZAWA Hiroyuki wrote:
> > On Fri, 25 Sep 2009 18:34:56 +1000
> > Nigel Cunningham <ncunningham@crca.org.au> wrote:
> > 
> >> Hi.
> >>
> >> KAMEZAWA Hiroyuki wrote:
> >>>> I have some code in TuxOnIce that needs a bit too (explicitly mark the
> >>>> VMA as needing to be atomically copied, for GEM objects), and am not
> >>>> sure what the canonical way to proceed is. Should a new unsigned long be
> >>>> added? The difficulty I see with that is that my flag was used in
> >>>> shmem_file_setup's flags parameter (drm_gem_object_alloc), so that
> >>>> function would need an extra parameter too..
> >>> Hmm, how about adding vma->vm_flags2 ?
> >> The difficulty there is that some functions pass these flags as arguments.
> >>
> > Ah yes. But I wonder some special flags, which is rarey used, can be moved
> > to vm_flags2...

But please don't call it vm_flags2. I think its better to partition the
existing flags by there purpose somehow hand give the flags fields
appropriate names which express that purpose.

	Joerg

Re: No more bits in vm_area_struct's vm_flags.

[Nigel Cunningham]

Hi.

KAMEZAWA Hiroyuki wrote:
> On Fri, 25 Sep 2009 18:34:56 +1000
> Nigel Cunningham <ncunningham@crca.org.au> wrote:
>> Hi.
>>
>> KAMEZAWA Hiroyuki wrote:
>>>> I have some code in TuxOnIce that needs a bit too (explicitly mark the
>>>> VMA as needing to be atomically copied, for GEM objects), and am not
>>>> sure what the canonical way to proceed is. Should a new unsigned long be
>>>> added? The difficulty I see with that is that my flag was used in
>>>> shmem_file_setup's flags parameter (drm_gem_object_alloc), so that
>>>> function would need an extra parameter too..
>>> Hmm, how about adding vma->vm_flags2 ?
>> The difficulty there is that some functions pass these flags as arguments.
>>
> Ah yes. But I wonder some special flags, which is rarey used, can be moved
> to vm_flags2...
> 
> For example,
> 
>  #define VM_SEQ_READ     0x00008000      /* App will access data sequentially */
>  #define VM_RAND_READ    0x00010000      /* App will not benefit from clustered reads */
> are all capsuled under
> mm.h
>  117 #define VM_READHINTMASK                 (VM_SEQ_READ | VM_RAND_READ)
>  118 #define VM_ClearReadHint(v)             (v)->vm_flags &= ~VM_READHINTMASK
>  119 #define VM_NormalReadHint(v)            (!((v)->vm_flags & VM_READHINTMASK))
>  120 #define VM_SequentialReadHint(v)        ((v)->vm_flags & VM_SEQ_READ)
>  121 #define VM_RandomReadHint(v)            ((v)->vm_flags & VM_RAND_READ)
> 
> Or
> 
> 105 #define VM_PFN_AT_MMAP  0x40000000      /* PFNMAP vma that is fully mapped at mmap time */
> is only used under special situation.
> 
> etc..
> 
> They'll be able to be moved to other(new) flag field, IIUC.

I'm working on a patch to do this, and am looking at is_mergeable_vma,
which is invoked via can_vma_merge_after from vma_merge from
madvise_behaviour (which potentially modifies these hint flags). Should
those hints be considered in that function? (Do I need to pass the hints
in as well and check they're equal?)

By the way, VM_ClearReadHint and VM_NormalReadHint are currently unused.
 madvise_behaviour manipulates the flags directly (in preparing
potential replacement values). Not sure if something should be done
about that.

By the way #2, in response to the later message in this thread, I'm
calling the new var vma->vm_hints, and have put it at the end of the
struct at the moment. Is that a good place?

Regards,

Nigel

Re: No more bits in vm_area_struct's vm_flags.

[KAMEZAWA Hiroyuki]

Hi,

On Mon, 28 Sep 2009 12:45:35 +1000
Nigel Cunningham <ncunningham@crca.org.au> wrote:
> KAMEZAWA Hiroyuki wrote:
> > On Fri, 25 Sep 2009 18:34:56 +1000
> > Nigel Cunningham <ncunningham@crca.org.au> wrote:
> >> Hi.
> >>
> >> KAMEZAWA Hiroyuki wrote:
> >>>> I have some code in TuxOnIce that needs a bit too (explicitly mark the
> >>>> VMA as needing to be atomically copied, for GEM objects), and am not
> >>>> sure what the canonical way to proceed is. Should a new unsigned long be
> >>>> added? The difficulty I see with that is that my flag was used in
> >>>> shmem_file_setup's flags parameter (drm_gem_object_alloc), so that
> >>>> function would need an extra parameter too..
> >>> Hmm, how about adding vma->vm_flags2 ?
> >> The difficulty there is that some functions pass these flags as arguments.
> >>
> > Ah yes. But I wonder some special flags, which is rarey used, can be moved
> > to vm_flags2...
> > 
> > For example,
> > 
> >  #define VM_SEQ_READ     0x00008000      /* App will access data sequentially */
> >  #define VM_RAND_READ    0x00010000      /* App will not benefit from clustered reads */
> > are all capsuled under
> > mm.h
> >  117 #define VM_READHINTMASK                 (VM_SEQ_READ | VM_RAND_READ)
> >  118 #define VM_ClearReadHint(v)             (v)->vm_flags &= ~VM_READHINTMASK
> >  119 #define VM_NormalReadHint(v)            (!((v)->vm_flags & VM_READHINTMASK))
> >  120 #define VM_SequentialReadHint(v)        ((v)->vm_flags & VM_SEQ_READ)
> >  121 #define VM_RandomReadHint(v)            ((v)->vm_flags & VM_RAND_READ)
> > 
> > Or
> > 
> > 105 #define VM_PFN_AT_MMAP  0x40000000      /* PFNMAP vma that is fully mapped at mmap time */
> > is only used under special situation.
> > 
> > etc..
> > 
> > They'll be able to be moved to other(new) flag field, IIUC.
> 
CCing Fengguang.

Breif Summary of thread:
 Now, vm->vm_flags has no more avialable bits. Then, I proposed Nigel
 to move some flags from vm->vm_flags to other flags as vm->vm_????.
 It seems readahead-hints are candidates for this......

> I'm working on a patch to do this, and am looking at is_mergeable_vma,
> which is invoked via can_vma_merge_after from vma_merge from
> madvise_behaviour (which potentially modifies these hint flags). Should
> those hints be considered in that function? (Do I need to pass the hints
> in as well and check they're equal?)

I think it should be handled.
But, yes it implies to add a new argument to several functions in mmap.c
and maybe a patch will be ugly.

How about addding this check ?

is_mergeable_vma(...)
....
  if (vma->vm_hints)
	return 0;

And not calling vma_merge() at madvice(ACCESS_PATTERN_HINT).

I wonder there are little chances when madice(ACCESS_PATTERN_HINT) is
given against mapped-file-vma...

> 
> By the way, VM_ClearReadHint and VM_NormalReadHint are currently unused.
>  madvise_behaviour manipulates the flags directly (in preparing
> potential replacement values). Not sure if something should be done
> about that.
> 
> By the way #2, in response to the later message in this thread, I'm
> calling the new var vma->vm_hints, and have put it at the end of the
> struct at the moment. Is that a good place?
> 
I think it seems nice. But please before entries under CONFIG.
How about just after vm_private_data ?

Regards,
-Kame

Re: No more bits in vm_area_struct's vm_flags.

[Wu Fengguang]

On Mon, Sep 28, 2009 at 11:04:50AM +0800, KAMEZAWA Hiroyuki wrote:
> Hi,
> 
> On Mon, 28 Sep 2009 12:45:35 +1000
> Nigel Cunningham <ncunningham@crca.org.au> wrote:
> > KAMEZAWA Hiroyuki wrote:
> > > On Fri, 25 Sep 2009 18:34:56 +1000
> > > Nigel Cunningham <ncunningham@crca.org.au> wrote:
> > >> Hi.
> > >>
> > >> KAMEZAWA Hiroyuki wrote:
> > >>>> I have some code in TuxOnIce that needs a bit too (explicitly mark the
> > >>>> VMA as needing to be atomically copied, for GEM objects), and am not
> > >>>> sure what the canonical way to proceed is. Should a new unsigned long be
> > >>>> added? The difficulty I see with that is that my flag was used in
> > >>>> shmem_file_setup's flags parameter (drm_gem_object_alloc), so that
> > >>>> function would need an extra parameter too..
> > >>> Hmm, how about adding vma->vm_flags2 ?
> > >> The difficulty there is that some functions pass these flags as arguments.
> > >>
> > > Ah yes. But I wonder some special flags, which is rarey used, can be moved
> > > to vm_flags2...
> > > 
> > > For example,
> > > 
> > >  #define VM_SEQ_READ     0x00008000      /* App will access data sequentially */
> > >  #define VM_RAND_READ    0x00010000      /* App will not benefit from clustered reads */
> > > are all capsuled under
> > > mm.h
> > >  117 #define VM_READHINTMASK                 (VM_SEQ_READ | VM_RAND_READ)
> > >  118 #define VM_ClearReadHint(v)             (v)->vm_flags &= ~VM_READHINTMASK
> > >  119 #define VM_NormalReadHint(v)            (!((v)->vm_flags & VM_READHINTMASK))
> > >  120 #define VM_SequentialReadHint(v)        ((v)->vm_flags & VM_SEQ_READ)
> > >  121 #define VM_RandomReadHint(v)            ((v)->vm_flags & VM_RAND_READ)
> > > 
> > > Or
> > > 
> > > 105 #define VM_PFN_AT_MMAP  0x40000000      /* PFNMAP vma that is fully mapped at mmap time */
> > > is only used under special situation.
> > > 
> > > etc..
> > > 
> > > They'll be able to be moved to other(new) flag field, IIUC.
> > 
> CCing Fengguang.
> 
> Breif Summary of thread:
>  Now, vm->vm_flags has no more avialable bits. Then, I proposed Nigel
>  to move some flags from vm->vm_flags to other flags as vm->vm_????.
>  It seems readahead-hints are candidates for this......

Agreed and thanks for the summary.

> > I'm working on a patch to do this, and am looking at is_mergeable_vma,
> > which is invoked via can_vma_merge_after from vma_merge from
> > madvise_behaviour (which potentially modifies these hint flags). Should
> > those hints be considered in that function? (Do I need to pass the hints
> > in as well and check they're equal?)
> 
> I think it should be handled.

Agreed.

> But, yes it implies to add a new argument to several functions in mmap.c
> and maybe a patch will be ugly.
> 
> How about addding this check ?
> 
> is_mergeable_vma(...)
> ....
>   if (vma->vm_hints)
> 	return 0;
> 
> And not calling vma_merge() at madvice(ACCESS_PATTERN_HINT).
> 
> I wonder there are little chances when madice(ACCESS_PATTERN_HINT) is
> given against mapped-file-vma...

Me wonder too. The access hints should be rarely used.
A simple solution is reasonable for them.

But what if more flags going into vm_hints in future?

Thanks,
Fengguang

> > 
> > By the way, VM_ClearReadHint and VM_NormalReadHint are currently unused.
> >  madvise_behaviour manipulates the flags directly (in preparing
> > potential replacement values). Not sure if something should be done
> > about that.
> > 
> > By the way #2, in response to the later message in this thread, I'm
> > calling the new var vma->vm_hints, and have put it at the end of the
> > struct at the moment. Is that a good place?
> > 
> I think it seems nice. But please before entries under CONFIG.
> How about just after vm_private_data ?
> 
> Regards,
> -Kame
> 

Re: No more bits in vm_area_struct's vm_flags.

[KAMEZAWA Hiroyuki]

On Mon, 28 Sep 2009 11:36:24 +0800
Wu Fengguang <fengguang.wu@intel.com> wrote:

> On Mon, Sep 28, 2009 at 11:04:50AM +0800, KAMEZAWA Hiroyuki wrote:
 
> > But, yes it implies to add a new argument to several functions in mmap.c
> > and maybe a patch will be ugly.
> > 
> > How about addding this check ?
> > 
> > is_mergeable_vma(...)
> > ....
> >   if (vma->vm_hints)
> > 	return 0;
> > 
> > And not calling vma_merge() at madvice(ACCESS_PATTERN_HINT).
> > 
> > I wonder there are little chances when madice(ACCESS_PATTERN_HINT) is
> > given against mapped-file-vma...
> 
> Me wonder too. The access hints should be rarely used.
> A simple solution is reasonable for them.
> 
> But what if more flags going into vm_hints in future?
> 

yes, that's concern...

But I wonder there are several flags which implies that the vma is not for
merge. as....

#define VM_GROWSDOWN    0x00000100      /* general info on the segment */
#define VM_GROWSUP      0x00000200

or some PFN_MAP flags...
(At least, we can have 4 flags of not-for-merge in vm_hints ;)

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 ;)

Thanks,
-Kame

Re: No more bits in vm_area_struct's vm_flags.

[Nigel Cunningham]

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.

Signed-off-by: Nigel Cunningham <nigel@tuxonice.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
 
+					/* 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;
 
 	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

Re: No more bits in vm_area_struct's vm_flags.

[Wu Fengguang]

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@intel.com>

> Signed-off-by: Nigel Cunningham <nigel@tuxonice.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

Re: No more bits in vm_area_struct's vm_flags.

[KAMEZAWA Hiroyuki]

On Mon, 28 Sep 2009 14:37:48 +1000
Nigel Cunningham <ncunningham@crca.org.au> 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.
> 
> Signed-off-by: Nigel Cunningham <nigel@tuxonice.net>

Seems good to me.

Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
But
> +	if (vma->vm_hints)
> +		return 0;
>  	return 1;

Maybe adding a comment (or more detailed patch description) is necessary.

Regards,
-Kame

>  }
>  
> -- 
> 1.6.3.3
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 

Re: No more bits in vm_area_struct's vm_flags.

[Nigel Cunningham]

Hi.

KAMEZAWA Hiroyuki wrote:
> Seems good to me.
> 
> Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
> But
>> +	if (vma->vm_hints)
>> +		return 0;
>>  	return 1;
> 
> Maybe adding a comment (or more detailed patch description) is necessary.

Thinking about this some more, I think we should also be looking at whether the new hints are non zero. Perhaps I should just add the new value to the
function parameters and be done with it.

Nigel

Re: No more bits in vm_area_struct's vm_flags.

[KAMEZAWA Hiroyuki]

On Mon, 28 Sep 2009 15:22:08 +1000
Nigel Cunningham <ncunningham@crca.org.au> wrote:

> Hi.
> 
> KAMEZAWA Hiroyuki wrote:
> > Seems good to me.
> > 
> > Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
> > But
> >> +	if (vma->vm_hints)
> >> +		return 0;
> >>  	return 1;
> > 
> > Maybe adding a comment (or more detailed patch description) is necessary.
> 
> Thinking about this some more, I think we should also be looking at whether the new hints are non zero. Perhaps I should just add the new value to the
> function parameters and be done with it.
> 
No objections from me. plz do.
I said option (1) just because patch size will be big to unexpected.
Thank you for your effort.

Regards,
-Kame

Re: No more bits in vm_area_struct's vm_flags.

[Hugh Dickins]

On Mon, 28 Sep 2009, KAMEZAWA Hiroyuki wrote:
> 
> What I dislike is making vm_flags to be long long ;)

Why?
Hugh

Re: No more bits in vm_area_struct's vm_flags.

[KAMEZAWA Hiroyuki]

Hugh Dickins wrote:
> On Mon, 28 Sep 2009, KAMEZAWA Hiroyuki wrote:
>>
>> What I dislike is making vm_flags to be long long ;)
>
> Why?
I'm sorry if my "dislike" sounds too strong.

Every time I see long long in the kernel, my concern is
"do I need spinlock to access this for 32bit arch ? is it safe ?".
(And it makes binary=>disassemble=>C (by eyes) a bit difficult)
Then, I don't like long long personally.

Another reason is some other calls like test_bit() cannot be used against
long long. (even if it's not used _now_)

Maybe vm->vm_flags will not require extra locks because
it can be protected by bigger lock as mmap_sem. Then, please make it
to be long long if its's recommended.

keeping vm_flags to be 32bit may makes vma_merge() ugly.
If so, long long is  a choice.

Thanks,
-Kame


> Hugh
>

Re: No more bits in vm_area_struct's vm_flags.

[Hugh Dickins]

On Tue, 29 Sep 2009, KAMEZAWA Hiroyuki wrote:
> Hugh Dickins wrote:
> > On Mon, 28 Sep 2009, KAMEZAWA Hiroyuki wrote:
> >>
> >> What I dislike is making vm_flags to be long long ;)
> >
> > Why?
> I'm sorry if my "dislike" sounds too strong.

No, not at all, I like the honesty with which you say "dislike".

> 
> Every time I see long long in the kernel, my concern is
> "do I need spinlock to access this for 32bit arch ? is it safe ?".
> (And it makes binary=>disassemble=>C (by eyes) a bit difficult)
> Then, I don't like long long personally.
> 
> Another reason is some other calls like test_bit() cannot be used against
> long long. (even if it's not used _now_)
> 
> Maybe vm->vm_flags will not require extra locks because
> it can be protected by bigger lock as mmap_sem.

I think that even as you wrote, you guessed I wouldn't be persuaded ;)
It sounds like you've had a bad experience with a long long in the past.

We already have to have locking for vm_flags, of course we do: it's
mmap_sem, yes, though I think you'll find some exceptions which know
they have exclusive access without it.

We use ordinary logical operations on vm_flags, we don't need it to
be atomic, we don't need an additional spinlock, we don't need to use
test_bit().  It's very easy!  (But irritating in a few places which
have to down_write mmap_sem for no other reason than to update vm_flags.)

> Then, please make it to be long long if its's recommended.
> 
> keeping vm_flags to be 32bit may makes vma_merge() ugly.
> If so, long long is  a choice.

unsigned long long is certainly the natural choice: that way leaves
freedom for people to add more flags in future without worrying about
which flags variable to put them into.  I'd better explain some of my
objections to Nigel's patch in a reply to him rather than here.

I have made up a patch to convert it to unsigned long long (not gone
through all the arches yet though), mainly to try a build to see how
it works out in practice.  I used a config which built most of the
non-debugging objects in mm/, things like migration and mempolicy
and memcg and ksm and so forth, but not kmemleak.

And I have to admit that the 834 bytes it added to i386 kernel text
is more than I was expecting, more than I can just brush away as "in
the noise".  I don't fully understand it yet.  There's a few silly
"andl $0xffffffff"s from the compiler (4.3.2), but not enough to
worry about.  Typically enlarged objects grow by 4 bytes, presumably
clearing the upper half when setting vma->vm_flags, fair enough.

300 bytes of growth is in mmap.o, 100 bytes of that in do_mmap_pgoff();
yet I don't see why it needed to grow by more than, say, 12 bytes.

My current feeling is that unsigned long long is the right way to
go, but given the bloat, we shouldn't convert over until we need to:
right now we should look to save a few flags instead.

Hugh

Re: No more bits in vm_area_struct's vm_flags.

[Hugh Dickins]

On Mon, 28 Sep 2009, Nigel Cunningham wrote:
> 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.

Arbitrary limitation, but not important at this stage.

> 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.
> 
> Signed-off-by: Nigel Cunningham <nigel@tuxonice.net>

This is quite a small patch (much smaller than my unsigned long long
one, not that I've completed or posted that yet); but I'm afraid that
I do not like it.

First off, do you realize that unsigned long vm_hints adds a pointless
8 bytes to every vm_area_struct on 64-bit systems?  Pointless because
there was already space for the flags you've moved in 64-bit vm_flags.

You might eliminate the bloat by making it an unsigned int or unsigned
char and squeezing it into a gap between other fields; but it still
seems silly to separate them.

I agree that VM_SEQ_READ and VM_RAND_READ are low hanging fruit,
in that they stand out as odd, partly because of the CamelCase
macros which hide them from sight.  But many others of the flags
are their own peculiar cases too.

I don't think there's a great queue of people wanting to add more
read hints: nobody would be more likely to do so than Fengguang,
and even he worries about "vm_hints" limiting its scope.

If you're looking to make more flagspace available, then it would
need to accomodate motley other flags too, so should be called...
vm_flags2?  shudder!

Were it not for the bloat, I wouldn't hesitate over unsigned long long.

As it is, I think that (as I said before) we should first look to see
if we can cull just a few of the flags which have accumulated.  Then
move on to unsigned long long when we have to.

I suggested before that for the moment you reuse VM_MAPPED_COPY,
and you said "Okee doke".  What changed?

Hugh

Re: No more bits in vm_area_struct's vm_flags.

[Nigel Cunningham]

Hi.

Hugh Dickins wrote:
> On Tue, 29 Sep 2009, KAMEZAWA Hiroyuki wrote:
> unsigned long long is certainly the natural choice: that way leaves
> freedom for people to add more flags in future without worrying about
> which flags variable to put them into.  I'd better explain some of my
> objections to Nigel's patch in a reply to him rather than here.

I'd prefer long long too if it's feasible. I'm just not an expert on the
issues, and so went in the direction I was pushed :) It looks a lot
cleaner and simpler to me, too, to keep everything in one variable.

Regards,

Nigel

Re: No more bits in vm_area_struct's vm_flags.

[Nigel Cunningham]

Hi Hugh.

Hugh Dickins wrote:
> I suggested before that for the moment you reuse VM_MAPPED_COPY,
> and you said "Okee doke".  What changed?

I made a mistake - got stuck at the point where I wasn't sure that was
an option. Everything had gone quiet and I wanted to make some headway,
so I started trying to make some progress by working on the path I
knew would be doable.

Sorry about that.

Nigel

Re: No more bits in vm_area_struct's vm_flags.

[KAMEZAWA Hiroyuki]

On Mon, 28 Sep 2009 22:00:20 +0100 (BST)
Hugh Dickins <hugh.dickins@tiscali.co.uk> wrote:

>> On Tue, 29 Sep 2009, KAMEZAWA Hiroyuki wrote:
> > 
> > Every time I see long long in the kernel, my concern is
> > "do I need spinlock to access this for 32bit arch ? is it safe ?".
> > (And it makes binary=>disassemble=>C (by eyes) a bit difficult)
> > Then, I don't like long long personally.
> > 
> > Another reason is some other calls like test_bit() cannot be used against
> > long long. (even if it's not used _now_)
> > 
> > Maybe vm->vm_flags will not require extra locks because
> > it can be protected by bigger lock as mmap_sem.
> 
> I think that even as you wrote, you guessed I wouldn't be persuaded ;)
> It sounds like you've had a bad experience with a long long in the past.
> 
yes ;)


> We already have to have locking for vm_flags, of course we do: it's
> mmap_sem, yes, though I think you'll find some exceptions which know
> they have exclusive access without it.
> 
> We use ordinary logical operations on vm_flags, we don't need it to
> be atomic, we don't need an additional spinlock, we don't need to use
> test_bit().  It's very easy!  (But irritating in a few places which
> have to down_write mmap_sem for no other reason than to update vm_flags.)
> 
Okay, I'll have no objections. 

Just a notice from lines stripped by grep

(1) using "int"  will be never correct even on 32bit.
==
vm_flags          242 arch/mips/mm/c-r3k.c 	int exec = vma->vm_flags & VM_EXEC;
vm_flags          293 drivers/char/mem.c 	return vma->vm_flags & VM_MAYSHARE;
vm_flags           44 mm/madvise.c   	int new_flags = vma->vm_flags;
vm_flags          547 mm/memory.c    	unsigned long vm_flags = vma->vm_flags;

But yes, it will be not a terrible bug for a while.

(2) All vm macros should be defined with ULL suffix. for supporing ~ 
==
vm_flags           30 arch/x86/mm/hugetlbpage.c 	unsigned long vm_flags = vma->vm_flags & ~VM_LOCKED;

(3) vma_merge()'s vm_flags should be ULL.


Not so many places as I thought..




> > Then, please make it to be long long if its's recommended.
> > 
> > keeping vm_flags to be 32bit may makes vma_merge() ugly.
> > If so, long long is  a choice.
> 
> unsigned long long is certainly the natural choice: that way leaves
> freedom for people to add more flags in future without worrying about
> which flags variable to put them into.  I'd better explain some of my
> objections to Nigel's patch in a reply to him rather than here.
> 
> I have made up a patch to convert it to unsigned long long (not gone
> through all the arches yet though), mainly to try a build to see how
> it works out in practice.  I used a config which built most of the
> non-debugging objects in mm/, things like migration and mempolicy
> and memcg and ksm and so forth, but not kmemleak.
> 
> And I have to admit that the 834 bytes it added to i386 kernel text
> is more than I was expecting, more than I can just brush away as "in
> the noise".  I don't fully understand it yet.  There's a few silly
> "andl $0xffffffff"s from the compiler (4.3.2), but not enough to
> worry about.  Typically enlarged objects grow by 4 bytes, presumably
> clearing the upper half when setting vma->vm_flags, fair enough.
> 
> 300 bytes of growth is in mmap.o, 100 bytes of that in do_mmap_pgoff();
> yet I don't see why it needed to grow by more than, say, 12 bytes.
> 
> My current feeling is that unsigned long long is the right way to
> go, but given the bloat, we shouldn't convert over until we need to:
> right now we should look to save a few flags instead.

Okay,

Thanks,
-Kame
 

Re: No more bits in vm_area_struct's vm_flags.

[Christoph Lameter]

Another concern that has not been discussed is the increased cache
footprint due to a slightly enlarged vm data working set (there is also a
corresponding icache issue since additional accesses are needed).

Could we stick with the current size and do combinations of flags like we
do with page flags? VM_HUGETLB cannot grow up and down f.e. and there are
certainly lots of other impossible combinations that can be used to put
more information into the flags.

Re: No more bits in vm_area_struct's vm_flags.

[Pavel Machek]

Hi!

> > Does TuxOnIce rely on CONFIG_MMU?  If so, then the TuxOnIce patch
> > could presumably reuse VM_MAPPED_COPY for now - but don't be
> > surprised if that's one we clean away later on.
> 
> Hmm. I'm not sure. The requirements are the same as for swsusp and
> uswsusp. Is there some tool to graph config dependencies?

I don't think swsusp was ported on any -nommu architecture, so config
dependency on MMU should be ok. OTOH such port should be doable...

								Pavel 

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

swsusp on nommu, was 'Re: No more bits in vm_area_struct's vm_flags.'

[Johannes Weiner]

On Wed, Sep 30, 2009 at 02:02:03PM +0200, Pavel Machek wrote:
> Hi!
> 
> > > Does TuxOnIce rely on CONFIG_MMU?  If so, then the TuxOnIce patch
> > > could presumably reuse VM_MAPPED_COPY for now - but don't be
> > > surprised if that's one we clean away later on.
> > 
> > Hmm. I'm not sure. The requirements are the same as for swsusp and
> > uswsusp. Is there some tool to graph config dependencies?
> 
> I don't think swsusp was ported on any -nommu architecture, so config
> dependency on MMU should be ok. OTOH such port should be doable...

I am sitting on some dusty patches to split swapfile handling from
actual paging and implement swsusp on blackfin.  They are incomplete
and I only occasionally find the time to continue working on them.  If
somebody is interested or also working on it, please let me know.

	Hannes

Re: swsusp on nommu, was 'Re: No more bits in vm_area_struct's vm_flags.'

[Mike Frysinger]

On Wed, Sep 30, 2009 at 09:08, Johannes Weiner wrote:
> On Wed, Sep 30, 2009 at 02:02:03PM +0200, Pavel Machek wrote:
>> > > Does TuxOnIce rely on CONFIG_MMU?  If so, then the TuxOnIce patch
>> > > could presumably reuse VM_MAPPED_COPY for now - but don't be
>> > > surprised if that's one we clean away later on.
>> >
>> > Hmm. I'm not sure. The requirements are the same as for swsusp and
>> > uswsusp. Is there some tool to graph config dependencies?
>>
>> I don't think swsusp was ported on any -nommu architecture, so config
>> dependency on MMU should be ok. OTOH such port should be doable...
>
> I am sitting on some dusty patches to split swapfile handling from
> actual paging and implement swsusp on blackfin.  They are incomplete
> and I only occasionally find the time to continue working on them.  If
> somebody is interested or also working on it, please let me know.

suspend to ram works on Blackfin systems w/out a problem, and i cant
think of a reason off the top of my head as to why saving/restoring
the image to disk couldnt work w/out a mmu ...

last time we looked, we found it too coupled to mmu code and we didnt
have a lot of pressure to get it done, so we moved on to other things.
-mike

Re: No more bits in vm_area_struct's vm_flags.

[Hugh Dickins]

On Tue, 29 Sep 2009, Christoph Lameter wrote:
> 
> Another concern that has not been discussed is the increased cache
> footprint due to a slightly enlarged vm data working set (there is also a
> corresponding icache issue since additional accesses are needed).

Using unsigned long long vm_flags makes no difference to cache footprint
on 64-bit systems, being a no-op there; and I think these days, though
we sure like our 32-bit systems to run well, we're not so anxious about
saving every last cycle on them.

> 
> Could we stick with the current size and do combinations of flags like we
> do with page flags?

Are we doing that?  If you have some example like, when PG_slab is set
then PG_owner_priv_1 means such-and-such, but if not not: okay, I'm
fine with that.

But if you're saying something like, if PG_reclaim is set at the same
time as PG_buddy, then they mean the page is not a buddy or under
reclaim, but brokenbacked: then I'm a bit (or even 32 bits) worried.

> VM_HUGETLB cannot grow up and down f.e. and there are
> certainly lots of other impossible combinations that can be used to put
> more information into the flags.

Where it makes sense, where it's understandable, okay: there may be a
few which could naturally use combinations.  But in general, no, I
think we'd be asking for endless maintenance trouble if we change the
meaning of some flags according to other flags.

Hugh

Re: No more bits in vm_area_struct's vm_flags.

[Hugh Dickins]

On Tue, 29 Sep 2009, KAMEZAWA Hiroyuki wrote:
> 
> (1) using "int"  will be never correct even on 32bit.
> ==
> vm_flags          242 arch/mips/mm/c-r3k.c 	int exec = vma->vm_flags & VM_EXEC;
> vm_flags          293 drivers/char/mem.c 	return vma->vm_flags & VM_MAYSHARE;
> vm_flags           44 mm/madvise.c   	int new_flags = vma->vm_flags;
> vm_flags          547 mm/memory.c    	unsigned long vm_flags = vma->vm_flags;
> 
> But yes, it will be not a terrible bug for a while.

There may be a few, probably recently added and rarely used, vm_flags
which we could consider moving from the low int to the high int of an
unsigned long long, for cheaper operations where they're not involved.

But I've no intention to shuffle the majority of vm_flags around:
VM_READ through VM_MAYSHARE will stay precisely where they are,
and a lot of others beyond e.g. I don't see any need to make mm->
def_flags an unsigned long long, so VM_LOCKED should stay low.

You say "using int will be never correct even on 32bit": you mean,
when dealing with an unknown set of vm_flags, as in your madvise.c
and memory.c examples above.  But the VM_EXEC example in arch/mips,
I don't see any need to pester an arch maintainer to change it.
The drivers/char/mem.c one I hadn't noticed (thanks), that one I'll
happily change: not because it represents a bug, but just to set a
good example to people adding such tests on higher flags in future.

> 
> (2) All vm macros should be defined with ULL suffix. for supporing ~ 
> ==
> vm_flags           30 arch/x86/mm/hugetlbpage.c 	unsigned long vm_flags = vma->vm_flags & ~VM_LOCKED;
> 
> (3) vma_merge()'s vm_flags should be ULL.

At first I thought you'd saved me a lot of embarrassment by mentioning
those ULL suffixes, I hadn't put them in.  But after a quick test of
what I thought was going to show a problem without them, no problem.
Please would you send me a test program which demonstrates the need
for all those ULLs?

Certainly the added higher flags would need ULLs to get through the
compiler (or preprocessor); and I don't mind adding ULLs throughout
as an example (though VM_EXEC will give ARM a slight problem), they
do seem to shave about 30 bytes off my build, presumably nudge the
compiler towards making better choices in a few places.  But I've
not yet seen why they're necessary.

I am *not* intending to pursue this further right now: I'd still
prefer us to be looking for vm_flags we can weed out.  But the day
is sure to come when we need to extend, so here for the record (and
for your curiosity!) is the patch I built up: complete so far as I
know, except for a VMA_VM_FLAGS endian question in ARM (I'm pretty
sure I'd get it wrong if I tried to adjust that one myself).

arch_mmap_check() was amusing: I should probably extract that
part and send it in as a separate patch, but no hurry.

I still don't understand why this adds 100 bytes to do_mmap_pgoff():
things I experimented with (e.g. changing vm_stat_account back to
taking unsigned long, changing the calc_vm algorithm back to not
be mixing types) only made a few bytes difference.

Hugh

 arch/arm/include/asm/cacheflush.h            |    7 -
 arch/arm/kernel/asm-offsets.c                |    3 
 arch/ia64/include/asm/mman.h                 |    5 
 arch/ia64/kernel/sys_ia64.c                  |    3 
 arch/powerpc/include/asm/mman.h              |    6 
 arch/s390/include/asm/mman.h                 |    2 
 arch/sh/mm/tlbflush_64.c                     |    2 
 arch/sparc/include/asm/mman.h                |    2 
 arch/x86/mm/hugetlbpage.c                    |    4 
 drivers/char/mem.c                           |    2 
 drivers/infiniband/hw/ipath/ipath_file_ops.c |    4 
 drivers/staging/android/binder.c             |    6 
 fs/binfmt_elf_fdpic.c                        |   12 -
 fs/exec.c                                    |    5 
 include/linux/hugetlb.h                      |    2 
 include/linux/ksm.h                          |    4 
 include/linux/mm.h                           |  109 +++++++----------
 include/linux/mm_types.h                     |    4 
 include/linux/mman.h                         |    8 -
 include/linux/rmap.h                         |    4 
 mm/filemap.c                                 |    6 
 mm/fremap.c                                  |    2 
 mm/ksm.c                                     |    2 
 mm/madvise.c                                 |    2 
 mm/memory.c                                  |   12 -
 mm/mlock.c                                   |   14 --
 mm/mmap.c                                    |   53 ++++----
 mm/mprotect.c                                |   12 -
 mm/mremap.c                                  |    2 
 mm/nommu.c                                   |   15 +-
 mm/rmap.c                                    |   10 -
 mm/shmem.c                                   |    2 
 mm/vmscan.c                                  |    4 
 33 files changed, 158 insertions(+), 172 deletions(-)

--- 2.6.32-rc1/arch/arm/include/asm/cacheflush.h	2009-09-28 00:27:55.000000000 +0100
+++ ull_vm_flags/arch/arm/include/asm/cacheflush.h	2009-09-29 16:48:15.000000000 +0100
@@ -169,7 +169,7 @@
  *		specified address space before a change of page tables.
  *		- start - user start address (inclusive, page aligned)
  *		- end   - user end address   (exclusive, page aligned)
- *		- flags - vma->vm_flags field
+ *		- flags - low unsigned long of vma->vm_flags field
  *
  *	coherent_kern_range(start, end)
  *
@@ -343,7 +343,7 @@ flush_cache_range(struct vm_area_struct
 {
 	if (cpumask_test_cpu(smp_processor_id(), mm_cpumask(vma->vm_mm)))
 		__cpuc_flush_user_range(start & PAGE_MASK, PAGE_ALIGN(end),
-					vma->vm_flags);
+					(unsigned long)vma->vm_flags);
 }
 
 static inline void
@@ -351,7 +351,8 @@ flush_cache_page(struct vm_area_struct *
 {
 	if (cpumask_test_cpu(smp_processor_id(), mm_cpumask(vma->vm_mm))) {
 		unsigned long addr = user_addr & PAGE_MASK;
-		__cpuc_flush_user_range(addr, addr + PAGE_SIZE, vma->vm_flags);
+		__cpuc_flush_user_range(addr, addr + PAGE_SIZE,
+					(unsigned long)vma->vm_flags);
 	}
 }
 
--- 2.6.32-rc1/arch/arm/kernel/asm-offsets.c	2008-07-13 22:51:29.000000000 +0100
+++ ull_vm_flags/arch/arm/kernel/asm-offsets.c	2009-09-29 16:48:15.000000000 +0100
@@ -88,8 +88,9 @@ int main(void)
 #endif
   DEFINE(VMA_VM_MM,		offsetof(struct vm_area_struct, vm_mm));
   DEFINE(VMA_VM_FLAGS,		offsetof(struct vm_area_struct, vm_flags));
+				/* but that will be wrong for bigendian? */
   BLANK();
-  DEFINE(VM_EXEC,	       	VM_EXEC);
+  DEFINE(VM_EXEC,	       	0x00000004);	/* mm.h now appends ULL */
   BLANK();
   DEFINE(PAGE_SZ,	       	PAGE_SIZE);
   BLANK();
--- 2.6.32-rc1/arch/ia64/include/asm/mman.h	2009-09-28 00:27:59.000000000 +0100
+++ ull_vm_flags/arch/ia64/include/asm/mman.h	2009-09-29 16:48:15.000000000 +0100
@@ -14,9 +14,8 @@
 
 #ifdef __KERNEL__
 #ifndef __ASSEMBLY__
-#define arch_mmap_check	ia64_mmap_check
-int ia64_mmap_check(unsigned long addr, unsigned long len,
-		unsigned long flags);
+#define arch_mmap_check(addr, len)	ia64_mmap_check(addr, len)
+int ia64_mmap_check(unsigned long addr, unsigned long len);
 #endif
 #endif
 
--- 2.6.32-rc1/arch/ia64/kernel/sys_ia64.c	2009-03-23 23:12:14.000000000 +0000
+++ ull_vm_flags/arch/ia64/kernel/sys_ia64.c	2009-09-29 16:48:15.000000000 +0100
@@ -169,8 +169,7 @@ sys_ia64_pipe (void)
 	return retval;
 }
 
-int ia64_mmap_check(unsigned long addr, unsigned long len,
-		unsigned long flags)
+int ia64_mmap_check(unsigned long addr, unsigned long len)
 {
 	unsigned long roff;
 
--- 2.6.32-rc1/arch/powerpc/include/asm/mman.h	2009-09-28 00:28:03.000000000 +0100
+++ ull_vm_flags/arch/powerpc/include/asm/mman.h	2009-09-29 16:48:15.000000000 +0100
@@ -38,13 +38,13 @@
  * This file is included by linux/mman.h, so we can't use cacl_vm_prot_bits()
  * here.  How important is the optimization?
  */
-static inline unsigned long arch_calc_vm_prot_bits(unsigned long prot)
+static inline unsigned long long arch_calc_vm_prot_bits(unsigned long prot)
 {
-	return (prot & PROT_SAO) ? VM_SAO : 0;
+	return (prot & PROT_SAO) ? VM_SAO : 0ULL;
 }
 #define arch_calc_vm_prot_bits(prot) arch_calc_vm_prot_bits(prot)
 
-static inline pgprot_t arch_vm_get_page_prot(unsigned long vm_flags)
+static inline pgprot_t arch_vm_get_page_prot(unsigned long long vm_flags)
 {
 	return (vm_flags & VM_SAO) ? __pgprot(_PAGE_SAO) : __pgprot(0);
 }
--- 2.6.32-rc1/arch/s390/include/asm/mman.h	2009-09-28 00:28:04.000000000 +0100
+++ ull_vm_flags/arch/s390/include/asm/mman.h	2009-09-29 16:48:15.000000000 +0100
@@ -12,8 +12,8 @@
 #include <asm-generic/mman.h>
 
 #if defined(__KERNEL__) && !defined(__ASSEMBLY__) && defined(CONFIG_64BIT)
+#define arch_mmap_check(addr, len)	s390_mmap_check(addr, len)
 int s390_mmap_check(unsigned long addr, unsigned long len);
-#define arch_mmap_check(addr,len,flags)	s390_mmap_check(addr,len)
 #endif
 
 #endif /* __S390_MMAN_H__ */
--- 2.6.32-rc1/arch/sh/mm/tlbflush_64.c	2009-09-28 00:28:11.000000000 +0100
+++ ull_vm_flags/arch/sh/mm/tlbflush_64.c	2009-09-29 16:48:15.000000000 +0100
@@ -48,7 +48,7 @@ static inline void print_vma(struct vm_a
 	printk("vma end   0x%08lx\n", vma->vm_end);
 
 	print_prots(vma->vm_page_prot);
-	printk("vm_flags 0x%08lx\n", vma->vm_flags);
+	printk("vm_flags 0x%08llx\n", vma->vm_flags);
 }
 
 static inline void print_task(struct task_struct *tsk)
--- 2.6.32-rc1/arch/sparc/include/asm/mman.h	2009-09-28 00:28:11.000000000 +0100
+++ ull_vm_flags/arch/sparc/include/asm/mman.h	2009-09-29 16:48:15.000000000 +0100
@@ -25,7 +25,7 @@
 
 #ifdef __KERNEL__
 #ifndef __ASSEMBLY__
-#define arch_mmap_check(addr,len,flags)	sparc_mmap_check(addr,len)
+#define arch_mmap_check(addr, len)	sparc_mmap_check(addr, len)
 int sparc_mmap_check(unsigned long addr, unsigned long len);
 #endif
 #endif
--- 2.6.32-rc1/arch/x86/mm/hugetlbpage.c	2009-06-10 04:05:27.000000000 +0100
+++ ull_vm_flags/arch/x86/mm/hugetlbpage.c	2009-09-29 16:48:15.000000000 +0100
@@ -27,8 +27,8 @@ static unsigned long page_table_shareabl
 	unsigned long s_end = sbase + PUD_SIZE;
 
 	/* Allow segments to share if only one is marked locked */
-	unsigned long vm_flags = vma->vm_flags & ~VM_LOCKED;
-	unsigned long svm_flags = svma->vm_flags & ~VM_LOCKED;
+	unsigned long long vm_flags = vma->vm_flags & ~VM_LOCKED;
+	unsigned long long svm_flags = svma->vm_flags & ~VM_LOCKED;
 
 	/*
 	 * match the virtual addresses, permission and the alignment of the
--- 2.6.32-rc1/drivers/char/mem.c	2009-09-28 00:28:15.000000000 +0100
+++ ull_vm_flags/drivers/char/mem.c	2009-09-29 16:48:15.000000000 +0100
@@ -290,7 +290,7 @@ static unsigned long get_unmapped_area_m
 /* can't do an in-place private mapping if there's no MMU */
 static inline int private_mapping_ok(struct vm_area_struct *vma)
 {
-	return vma->vm_flags & VM_MAYSHARE;
+	return (vma->vm_flags & VM_MAYSHARE) != 0;
 }
 #else
 #define get_unmapped_area_mem	NULL
--- 2.6.32-rc1/drivers/infiniband/hw/ipath/ipath_file_ops.c	2009-09-28 00:28:17.000000000 +0100
+++ ull_vm_flags/drivers/infiniband/hw/ipath/ipath_file_ops.c	2009-09-29 16:48:15.000000000 +0100
@@ -1112,7 +1112,7 @@ static int mmap_rcvegrbufs(struct vm_are
 
 	if (vma->vm_flags & VM_WRITE) {
 		dev_info(&dd->pcidev->dev, "Can't map eager buffers as "
-			 "writable (flags=%lx)\n", vma->vm_flags);
+			 "writable (flags=%llx)\n", vma->vm_flags);
 		ret = -EPERM;
 		goto bail;
 	}
@@ -1201,7 +1201,7 @@ static int mmap_kvaddr(struct vm_area_st
                 if (vma->vm_flags & VM_WRITE) {
                         dev_info(&dd->pcidev->dev,
                                  "Can't map eager buffers as "
-                                 "writable (flags=%lx)\n", vma->vm_flags);
+                                 "writable (flags=%llx)\n", vma->vm_flags);
                         ret = -EPERM;
                         goto bail;
                 }
--- 2.6.32-rc1/drivers/staging/android/binder.c	2009-09-28 00:28:28.000000000 +0100
+++ ull_vm_flags/drivers/staging/android/binder.c	2009-09-29 16:48:15.000000000 +0100
@@ -2737,7 +2737,7 @@ static void binder_vma_open(struct vm_ar
 {
 	struct binder_proc *proc = vma->vm_private_data;
 	binder_debug(BINDER_DEBUG_OPEN_CLOSE,
-		     "binder: %d open vm area %lx-%lx (%ld K) vma %lx pagep %lx\n",
+		     "binder: %d open vm area %lx-%lx (%ld K) vma %llx pagep %lx\n",
 		     proc->pid, vma->vm_start, vma->vm_end,
 		     (vma->vm_end - vma->vm_start) / SZ_1K, vma->vm_flags,
 		     (unsigned long)pgprot_val(vma->vm_page_prot));
@@ -2748,7 +2748,7 @@ static void binder_vma_close(struct vm_a
 {
 	struct binder_proc *proc = vma->vm_private_data;
 	binder_debug(BINDER_DEBUG_OPEN_CLOSE,
-		     "binder: %d close vm area %lx-%lx (%ld K) vma %lx pagep %lx\n",
+		     "binder: %d close vm area %lx-%lx (%ld K) vma %llx pagep %lx\n",
 		     proc->pid, vma->vm_start, vma->vm_end,
 		     (vma->vm_end - vma->vm_start) / SZ_1K, vma->vm_flags,
 		     (unsigned long)pgprot_val(vma->vm_page_prot));
@@ -2773,7 +2773,7 @@ static int binder_mmap(struct file *filp
 		vma->vm_end = vma->vm_start + SZ_4M;
 
 	binder_debug(BINDER_DEBUG_OPEN_CLOSE,
-		     "binder_mmap: %d %lx-%lx (%ld K) vma %lx pagep %lx\n",
+		     "binder_mmap: %d %lx-%lx (%ld K) vma %llx pagep %lx\n",
 		     proc->pid, vma->vm_start, vma->vm_end,
 		     (vma->vm_end - vma->vm_start) / SZ_1K, vma->vm_flags,
 		     (unsigned long)pgprot_val(vma->vm_page_prot));
--- 2.6.32-rc1/fs/binfmt_elf_fdpic.c	2009-09-28 00:28:34.000000000 +0100
+++ ull_vm_flags/fs/binfmt_elf_fdpic.c	2009-09-29 16:48:15.000000000 +0100
@@ -1235,7 +1235,7 @@ static int maydump(struct vm_area_struct
 
 	/* Do not dump I/O mapped devices or special mappings */
 	if (vma->vm_flags & (VM_IO | VM_RESERVED)) {
-		kdcore("%08lx: %08lx: no (IO)", vma->vm_start, vma->vm_flags);
+		kdcore("%08lx: %08llx: no (IO)", vma->vm_start, vma->vm_flags);
 		return 0;
 	}
 
@@ -1243,7 +1243,7 @@ static int maydump(struct vm_area_struct
 	 * them either. "dump_write()" can't handle it anyway.
 	 */
 	if (!(vma->vm_flags & VM_READ)) {
-		kdcore("%08lx: %08lx: no (!read)", vma->vm_start, vma->vm_flags);
+		kdcore("%08lx: %08llx: no (!read)", vma->vm_start, vma->vm_flags);
 		return 0;
 	}
 
@@ -1251,13 +1251,13 @@ static int maydump(struct vm_area_struct
 	if (vma->vm_flags & VM_SHARED) {
 		if (vma->vm_file->f_path.dentry->d_inode->i_nlink == 0) {
 			dump_ok = test_bit(MMF_DUMP_ANON_SHARED, &mm_flags);
-			kdcore("%08lx: %08lx: %s (share)", vma->vm_start,
+			kdcore("%08lx: %08llx: %s (share)", vma->vm_start,
 			       vma->vm_flags, dump_ok ? "yes" : "no");
 			return dump_ok;
 		}
 
 		dump_ok = test_bit(MMF_DUMP_MAPPED_SHARED, &mm_flags);
-		kdcore("%08lx: %08lx: %s (share)", vma->vm_start,
+		kdcore("%08lx: %08llx: %s (share)", vma->vm_start,
 		       vma->vm_flags, dump_ok ? "yes" : "no");
 		return dump_ok;
 	}
@@ -1266,14 +1266,14 @@ static int maydump(struct vm_area_struct
 	/* By default, if it hasn't been written to, don't write it out */
 	if (!vma->anon_vma) {
 		dump_ok = test_bit(MMF_DUMP_MAPPED_PRIVATE, &mm_flags);
-		kdcore("%08lx: %08lx: %s (!anon)", vma->vm_start,
+		kdcore("%08lx: %08llx: %s (!anon)", vma->vm_start,
 		       vma->vm_flags, dump_ok ? "yes" : "no");
 		return dump_ok;
 	}
 #endif
 
 	dump_ok = test_bit(MMF_DUMP_ANON_PRIVATE, &mm_flags);
-	kdcore("%08lx: %08lx: %s", vma->vm_start, vma->vm_flags,
+	kdcore("%08lx: %08llx: %s", vma->vm_start, vma->vm_flags,
 	       dump_ok ? "yes" : "no");
 	return dump_ok;
 }
--- 2.6.32-rc1/fs/exec.c	2009-09-28 00:28:35.000000000 +0100
+++ ull_vm_flags/fs/exec.c	2009-09-29 16:48:15.000000000 +0100
@@ -570,7 +570,7 @@ int setup_arg_pages(struct linux_binprm
 	struct mm_struct *mm = current->mm;
 	struct vm_area_struct *vma = bprm->vma;
 	struct vm_area_struct *prev = NULL;
-	unsigned long vm_flags;
+	unsigned long long vm_flags;
 	unsigned long stack_base;
 
 #ifdef CONFIG_STACK_GROWSUP
@@ -615,8 +615,7 @@ int setup_arg_pages(struct linux_binprm
 		vm_flags &= ~VM_EXEC;
 	vm_flags |= mm->def_flags;
 
-	ret = mprotect_fixup(vma, &prev, vma->vm_start, vma->vm_end,
-			vm_flags);
+	ret = mprotect_fixup(vma, &prev, vma->vm_start, vma->vm_end, vm_flags);
 	if (ret)
 		goto out_unlock;
 	BUG_ON(prev != vma);
--- 2.6.32-rc1/include/linux/hugetlb.h	2009-09-28 00:28:38.000000000 +0100
+++ ull_vm_flags/include/linux/hugetlb.h	2009-09-29 16:48:15.000000000 +0100
@@ -16,7 +16,7 @@ int PageHuge(struct page *page);
 
 static inline int is_vm_hugetlb_page(struct vm_area_struct *vma)
 {
-	return vma->vm_flags & VM_HUGETLB;
+	return (vma->vm_flags & VM_HUGETLB) != 0;
 }
 
 void reset_vma_resv_huge_pages(struct vm_area_struct *vma);
--- 2.6.32-rc1/include/linux/ksm.h	2009-09-28 00:28:38.000000000 +0100
+++ ull_vm_flags/include/linux/ksm.h	2009-09-29 16:48:15.000000000 +0100
@@ -14,7 +14,7 @@
 
 #ifdef CONFIG_KSM
 int ksm_madvise(struct vm_area_struct *vma, unsigned long start,
-		unsigned long end, int advice, unsigned long *vm_flags);
+		unsigned long end, int advice, unsigned long long *vm_flags);
 int __ksm_enter(struct mm_struct *mm);
 void __ksm_exit(struct mm_struct *mm);
 
@@ -54,7 +54,7 @@ static inline void page_add_ksm_rmap(str
 #else  /* !CONFIG_KSM */
 
 static inline int ksm_madvise(struct vm_area_struct *vma, unsigned long start,
-		unsigned long end, int advice, unsigned long *vm_flags)
+		unsigned long end, int advice, unsigned long long *vm_flags)
 {
 	return 0;
 }
--- 2.6.32-rc1/include/linux/mm.h	2009-09-28 00:28:38.000000000 +0100
+++ ull_vm_flags/include/linux/mm.h	2009-09-29 16:48:15.000000000 +0100
@@ -65,46 +65,49 @@ extern unsigned int kobjsize(const void
 /*
  * vm_flags in vm_area_struct, see mm_types.h.
  */
-#define VM_READ		0x00000001	/* currently active flags */
-#define VM_WRITE	0x00000002
-#define VM_EXEC		0x00000004
-#define VM_SHARED	0x00000008
+#define VM_READ		0x00000001ULL	/* currently active flags */
+#define VM_WRITE	0x00000002ULL
+#define VM_EXEC		0x00000004ULL
+#define VM_SHARED	0x00000008ULL
 
 /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
-#define VM_MAYREAD	0x00000010	/* limits for mprotect() etc */
-#define VM_MAYWRITE	0x00000020
-#define VM_MAYEXEC	0x00000040
-#define VM_MAYSHARE	0x00000080
-
-#define VM_GROWSDOWN	0x00000100	/* general info on the segment */
-#define VM_GROWSUP	0x00000200
-#define VM_PFNMAP	0x00000400	/* Page-ranges managed without "struct page", just pure PFN */
-#define VM_DENYWRITE	0x00000800	/* ETXTBSY on write attempts.. */
-
-#define VM_EXECUTABLE	0x00001000
-#define VM_LOCKED	0x00002000
-#define VM_IO           0x00004000	/* Memory mapped I/O or similar */
+#define VM_MAYREAD	0x00000010ULL	/* limits for mprotect() etc */
+#define VM_MAYWRITE	0x00000020ULL
+#define VM_MAYEXEC	0x00000040ULL
+#define VM_MAYSHARE	0x00000080ULL
+
+#define VM_GROWSDOWN	0x00000100ULL	/* general info on the segment */
+#define VM_GROWSUP	0x00000200ULL
+#define VM_PFNMAP	0x00000400ULL	/* Page-ranges managed without "struct page", just pure PFN */
+#define VM_DENYWRITE	0x00000800ULL	/* ETXTBSY on write attempts.. */
+
+#define VM_EXECUTABLE	0x00001000ULL
+#define VM_LOCKED	0x00002000ULL
+#define VM_IO   	0x00004000ULL	/* 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_SEQ_READ	0x00008000ULL	/* App will access data sequentially */
+#define VM_RAND_READ	0x00010000ULL	/* 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 */
-#define VM_ACCOUNT	0x00100000	/* Is a VM accounted object */
-#define VM_NORESERVE	0x00200000	/* should the VM suppress accounting */
-#define VM_HUGETLB	0x00400000	/* Huge TLB Page VM */
-#define VM_NONLINEAR	0x00800000	/* Is non-linear (remap_file_pages) */
-#define VM_MAPPED_COPY	0x01000000	/* T if mapped copy of data (nommu mmap) */
-#define VM_INSERTPAGE	0x02000000	/* The vma has had "vm_insert_page()" done on it */
-#define VM_ALWAYSDUMP	0x04000000	/* Always include in core dumps */
-
-#define VM_CAN_NONLINEAR 0x08000000	/* Has ->fault & does nonlinear pages */
-#define VM_MIXEDMAP	0x10000000	/* Can contain "struct page" and pure PFN pages */
-#define VM_SAO		0x20000000	/* Strong Access Ordering (powerpc) */
-#define VM_PFN_AT_MMAP	0x40000000	/* PFNMAP vma that is fully mapped at mmap time */
-#define VM_MERGEABLE	0x80000000	/* KSM may merge identical pages */
+#define VM_DONTCOPY	0x00020000ULL	/* Do not copy this vma on fork */
+#define VM_DONTEXPAND	0x00040000ULL	/* Cannot expand with mremap() */
+#define VM_RESERVED	0x00080000ULL	/* Count as reserved_vm like IO */
+
+#define VM_ACCOUNT	0x00100000ULL	/* Is a VM accounted object */
+#define VM_NORESERVE	0x00200000ULL	/* should the VM suppress accounting */
+#define VM_HUGETLB	0x00400000ULL	/* Huge TLB Page VM */
+#define VM_NONLINEAR	0x00800000ULL	/* Is non-linear (remap_file_pages) */
+
+#define VM_MAPPED_COPY	0x01000000ULL	/* T if mapped copy of data (nommu mmap) */
+#define VM_INSERTPAGE	0x02000000ULL	/* The vma has had "vm_insert_page()" done on it */
+#define VM_ALWAYSDUMP	0x04000000ULL	/* Always include in core dumps */
+
+#define VM_CAN_NONLINEAR 0x08000000ULL	/* Has ->fault & does nonlinear pages */
+
+#define VM_MIXEDMAP	0x10000000ULL	/* Can contain "struct page" and pure PFN pages */
+#define VM_SAO		0x20000000ULL	/* Strong Access Ordering (powerpc) */
+#define VM_PFN_AT_MMAP	0x40000000ULL	/* PFNMAP vma that is fully mapped at mmap time */
+#define VM_MERGEABLE	0x80000000ULL	/* KSM may merge identical pages */
 
 #ifndef VM_STACK_DEFAULT_FLAGS		/* arch can override this */
 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
@@ -116,12 +119,6 @@ extern unsigned int kobjsize(const void
 #define VM_STACK_FLAGS	(VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
 #endif
 
-#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)
-
 /*
  * special vmas that are non-mergable, non-mlock()able
  */
@@ -147,12 +144,12 @@ extern pgprot_t protection_map[16];
  */
 static inline int is_linear_pfn_mapping(struct vm_area_struct *vma)
 {
-	return (vma->vm_flags & VM_PFN_AT_MMAP);
+	return (vma->vm_flags & VM_PFN_AT_MMAP) != 0;
 }
 
 static inline int is_pfn_mapping(struct vm_area_struct *vma)
 {
-	return (vma->vm_flags & VM_PFNMAP);
+	return (vma->vm_flags & VM_PFNMAP) != 0;
 }
 
 /*
@@ -715,14 +712,6 @@ int shmem_lock(struct file *file, int lo
 struct file *shmem_file_setup(const char *name, loff_t size, unsigned long flags);
 int shmem_zero_setup(struct vm_area_struct *);
 
-#ifndef CONFIG_MMU
-extern unsigned long shmem_get_unmapped_area(struct file *file,
-					     unsigned long addr,
-					     unsigned long len,
-					     unsigned long pgoff,
-					     unsigned long flags);
-#endif
-
 extern int can_do_mlock(void);
 extern int user_shm_lock(size_t, struct user_struct *);
 extern void user_shm_unlock(size_t, struct user_struct *);
@@ -845,7 +834,7 @@ extern unsigned long do_mremap(unsigned
 			       unsigned long flags, unsigned long new_addr);
 extern int mprotect_fixup(struct vm_area_struct *vma,
 			  struct vm_area_struct **pprev, unsigned long start,
-			  unsigned long end, unsigned long newflags);
+			  unsigned long end, unsigned long long newflags);
 
 /*
  * doesn't attempt to fault and will return short.
@@ -1096,7 +1085,7 @@ extern void vma_adjust(struct vm_area_st
 	unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
 extern struct vm_area_struct *vma_merge(struct mm_struct *,
 	struct vm_area_struct *prev, unsigned long addr, unsigned long end,
-	unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
+	unsigned long long vm_flags, struct anon_vma *, struct file *, pgoff_t,
 	struct mempolicy *);
 extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
 extern int split_vma(struct mm_struct *,
@@ -1125,9 +1114,8 @@ static inline void removed_exe_file_vma(
 #endif /* CONFIG_PROC_FS */
 
 extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
-extern int install_special_mapping(struct mm_struct *mm,
-				   unsigned long addr, unsigned long len,
-				   unsigned long flags, struct page **pages);
+extern int install_special_mapping(struct mm_struct *mm, unsigned long addr,
+	unsigned long len, unsigned long long vm_flags, struct page **pages);
 
 extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
 
@@ -1136,7 +1124,7 @@ extern unsigned long do_mmap_pgoff(struc
 	unsigned long flag, unsigned long pgoff);
 extern unsigned long mmap_region(struct file *file, unsigned long addr,
 	unsigned long len, unsigned long flags,
-	unsigned int vm_flags, unsigned long pgoff);
+	unsigned long long vm_flags, unsigned long pgoff);
 
 static inline unsigned long do_mmap(struct file *file, unsigned long addr,
 	unsigned long len, unsigned long prot,
@@ -1222,7 +1210,7 @@ static inline unsigned long vma_pages(st
 	return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
 }
 
-pgprot_t vm_get_page_prot(unsigned long vm_flags);
+pgprot_t vm_get_page_prot(unsigned long long vm_flags);
 struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
 int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
 			unsigned long pfn, unsigned long size, pgprot_t);
@@ -1246,10 +1234,11 @@ extern int apply_to_page_range(struct mm
 			       unsigned long size, pte_fn_t fn, void *data);
 
 #ifdef CONFIG_PROC_FS
-void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
+void vm_stat_account(struct mm_struct *mm,
+		unsigned long long vm_flags, struct file *file, long pages);
 #else
 static inline void vm_stat_account(struct mm_struct *mm,
-			unsigned long flags, struct file *file, long pages)
+		unsigned long long vm_flags, struct file *file, long pages)
 {
 }
 #endif /* CONFIG_PROC_FS */
--- 2.6.32-rc1/include/linux/mm_types.h	2009-09-28 00:28:38.000000000 +0100
+++ ull_vm_flags/include/linux/mm_types.h	2009-09-29 16:48:15.000000000 +0100
@@ -115,7 +115,7 @@ struct page {
  */
 struct vm_region {
 	struct rb_node	vm_rb;		/* link in global region tree */
-	unsigned long	vm_flags;	/* VMA vm_flags */
+	unsigned long long vm_flags;	/* VMA vm_flags */
 	unsigned long	vm_start;	/* start address of region */
 	unsigned long	vm_end;		/* region initialised to here */
 	unsigned long	vm_top;		/* region allocated to here */
@@ -141,7 +141,7 @@ struct vm_area_struct {
 	struct vm_area_struct *vm_next;
 
 	pgprot_t vm_page_prot;		/* Access permissions of this VMA. */
-	unsigned long vm_flags;		/* Flags, see mm.h. */
+	unsigned long long vm_flags;	/* Flags, see mm.h. */
 
 	struct rb_node vm_rb;
 
--- 2.6.32-rc1/include/linux/mman.h	2009-06-10 04:05:27.000000000 +0100
+++ ull_vm_flags/include/linux/mman.h	2009-09-29 16:48:15.000000000 +0100
@@ -35,7 +35,7 @@ static inline void vm_unacct_memory(long
  */
 
 #ifndef arch_calc_vm_prot_bits
-#define arch_calc_vm_prot_bits(prot) 0
+#define arch_calc_vm_prot_bits(prot)	0ULL
 #endif
 
 #ifndef arch_vm_get_page_prot
@@ -69,8 +69,7 @@ static inline int arch_validate_prot(uns
 /*
  * Combine the mmap "prot" argument into "vm_flags" used internally.
  */
-static inline unsigned long
-calc_vm_prot_bits(unsigned long prot)
+static inline unsigned long long calc_vm_prot_bits(unsigned long prot)
 {
 	return _calc_vm_trans(prot, PROT_READ,  VM_READ ) |
 	       _calc_vm_trans(prot, PROT_WRITE, VM_WRITE) |
@@ -81,8 +80,7 @@ calc_vm_prot_bits(unsigned long prot)
 /*
  * Combine the mmap "flags" argument into "vm_flags" used internally.
  */
-static inline unsigned long
-calc_vm_flag_bits(unsigned long flags)
+static inline unsigned long long calc_vm_flag_bits(unsigned long flags)
 {
 	return _calc_vm_trans(flags, MAP_GROWSDOWN,  VM_GROWSDOWN ) |
 	       _calc_vm_trans(flags, MAP_DENYWRITE,  VM_DENYWRITE ) |
--- 2.6.32-rc1/include/linux/rmap.h	2009-09-28 00:28:39.000000000 +0100
+++ ull_vm_flags/include/linux/rmap.h	2009-09-29 16:48:15.000000000 +0100
@@ -80,7 +80,7 @@ static inline void page_dup_rmap(struct
  * Called from mm/vmscan.c to handle paging out
  */
 int page_referenced(struct page *, int is_locked,
-			struct mem_cgroup *cnt, unsigned long *vm_flags);
+			struct mem_cgroup *cnt, unsigned long long *vm_flags);
 enum ttu_flags {
 	TTU_UNMAP = 0,			/* unmap mode */
 	TTU_MIGRATION = 1,		/* migration mode */
@@ -135,7 +135,7 @@ int page_mapped_in_vma(struct page *page
 
 static inline int page_referenced(struct page *page, int is_locked,
 				  struct mem_cgroup *cnt,
-				  unsigned long *vm_flags)
+				  unsigned long long *vm_flags)
 {
 	*vm_flags = 0;
 	return TestClearPageReferenced(page);
--- 2.6.32-rc1/mm/filemap.c	2009-09-28 00:28:41.000000000 +0100
+++ ull_vm_flags/mm/filemap.c	2009-09-29 16:48:15.000000000 +0100
@@ -1429,10 +1429,10 @@ static void do_sync_mmap_readahead(struc
 	struct address_space *mapping = file->f_mapping;
 
 	/* If we don't want any read-ahead, don't bother */
-	if (VM_RandomReadHint(vma))
+	if (vma->vm_flags & VM_RAND_READ)
 		return;
 
-	if (VM_SequentialReadHint(vma) ||
+	if ((vma->vm_flags & VM_SEQ_READ) ||
 			offset - 1 == (ra->prev_pos >> PAGE_CACHE_SHIFT)) {
 		page_cache_sync_readahead(mapping, ra, file, offset,
 					  ra->ra_pages);
@@ -1474,7 +1474,7 @@ static void do_async_mmap_readahead(stru
 	struct address_space *mapping = file->f_mapping;
 
 	/* If we don't want any read-ahead, don't bother */
-	if (VM_RandomReadHint(vma))
+	if (vma->vm_flags & VM_RAND_READ)
 		return;
 	if (ra->mmap_miss > 0)
 		ra->mmap_miss--;
--- 2.6.32-rc1/mm/fremap.c	2009-03-23 23:12:14.000000000 +0000
+++ ull_vm_flags/mm/fremap.c	2009-09-29 16:48:15.000000000 +0100
@@ -221,7 +221,7 @@ SYSCALL_DEFINE5(remap_file_pages, unsign
 		/*
 		 * drop PG_Mlocked flag for over-mapped range
 		 */
-		unsigned int saved_flags = vma->vm_flags;
+		unsigned long long saved_flags = vma->vm_flags;
 		munlock_vma_pages_range(vma, start, start + size);
 		vma->vm_flags = saved_flags;
 	}
--- 2.6.32-rc1/mm/ksm.c	2009-09-28 00:28:41.000000000 +0100
+++ ull_vm_flags/mm/ksm.c	2009-09-29 16:48:15.000000000 +0100
@@ -1366,7 +1366,7 @@ static int ksm_scan_thread(void *nothing
 }
 
 int ksm_madvise(struct vm_area_struct *vma, unsigned long start,
-		unsigned long end, int advice, unsigned long *vm_flags)
+		unsigned long end, int advice, unsigned long long *vm_flags)
 {
 	struct mm_struct *mm = vma->vm_mm;
 	int err;
--- 2.6.32-rc1/mm/madvise.c	2009-09-28 00:28:41.000000000 +0100
+++ ull_vm_flags/mm/madvise.c	2009-09-29 16:48:15.000000000 +0100
@@ -42,7 +42,7 @@ static long madvise_behavior(struct vm_a
 	struct mm_struct * mm = vma->vm_mm;
 	int error = 0;
 	pgoff_t pgoff;
-	unsigned long new_flags = vma->vm_flags;
+	unsigned long long new_flags = vma->vm_flags;
 
 	switch (behavior) {
 	case MADV_NORMAL:
--- 2.6.32-rc1/mm/memory.c	2009-09-28 00:28:41.000000000 +0100
+++ ull_vm_flags/mm/memory.c	2009-09-29 16:48:15.000000000 +0100
@@ -437,7 +437,7 @@ static void print_bad_pte(struct vm_area
 		page_mapcount(page), page->mapping, page->index);
 	}
 	printk(KERN_ALERT
-		"addr:%p vm_flags:%08lx anon_vma:%p mapping:%p index:%lx\n",
+		"addr:%p vm_flags:%08llx anon_vma:%p mapping:%p index:%lx\n",
 		(void *)addr, vma->vm_flags, vma->anon_vma, mapping, index);
 	/*
 	 * Choose text because data symbols depend on CONFIG_KALLSYMS_ALL=y
@@ -452,9 +452,9 @@ static void print_bad_pte(struct vm_area
 	add_taint(TAINT_BAD_PAGE);
 }
 
-static inline int is_cow_mapping(unsigned int flags)
+static inline int is_cow_mapping(unsigned long long vm_flags)
 {
-	return (flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE;
+	return (vm_flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE;
 }
 
 #ifndef is_zero_pfn
@@ -577,7 +577,7 @@ copy_one_pte(struct mm_struct *dst_mm, s
 		pte_t *dst_pte, pte_t *src_pte, struct vm_area_struct *vma,
 		unsigned long addr, int *rss)
 {
-	unsigned long vm_flags = vma->vm_flags;
+	unsigned long long vm_flags = vma->vm_flags;
 	pte_t pte = *src_pte;
 	struct page *page;
 
@@ -852,7 +852,7 @@ static unsigned long zap_pte_range(struc
 				if (pte_dirty(ptent))
 					set_page_dirty(page);
 				if (pte_young(ptent) &&
-				    likely(!VM_SequentialReadHint(vma)))
+				    likely((!(vma->vm_flags & VM_SEQ_READ))))
 					mark_page_accessed(page);
 				file_rss--;
 			}
@@ -1231,7 +1231,7 @@ int __get_user_pages(struct task_struct
 		     struct page **pages, struct vm_area_struct **vmas)
 {
 	int i;
-	unsigned long vm_flags;
+	unsigned long long vm_flags;
 
 	if (nr_pages <= 0)
 		return 0;
--- 2.6.32-rc1/mm/mlock.c	2009-09-28 00:28:41.000000000 +0100
+++ ull_vm_flags/mm/mlock.c	2009-09-29 16:48:15.000000000 +0100
@@ -266,8 +266,7 @@ long mlock_vma_pages_range(struct vm_are
 	if (vma->vm_flags & (VM_IO | VM_PFNMAP))
 		goto no_mlock;
 
-	if (!((vma->vm_flags & (VM_DONTEXPAND | VM_RESERVED)) ||
-			is_vm_hugetlb_page(vma) ||
+	if (!((vma->vm_flags & (VM_DONTEXPAND | VM_RESERVED | VM_HUGETLB)) ||
 			vma == get_gate_vma(current))) {
 
 		__mlock_vma_pages_range(vma, start, end);
@@ -354,20 +353,19 @@ void munlock_vma_pages_range(struct vm_a
  * For vmas that pass the filters, merge/split as appropriate.
  */
 static int mlock_fixup(struct vm_area_struct *vma, struct vm_area_struct **prev,
-	unsigned long start, unsigned long end, unsigned int newflags)
+	unsigned long start, unsigned long end, unsigned long long newflags)
 {
 	struct mm_struct *mm = vma->vm_mm;
 	pgoff_t pgoff;
 	int nr_pages;
 	int ret = 0;
-	int lock = newflags & VM_LOCKED;
+	int lock = !!(newflags & VM_LOCKED);
 
 	if (newflags == vma->vm_flags ||
 			(vma->vm_flags & (VM_IO | VM_PFNMAP)))
 		goto out;	/* don't set VM_LOCKED,  don't count */
 
-	if ((vma->vm_flags & (VM_DONTEXPAND | VM_RESERVED)) ||
-			is_vm_hugetlb_page(vma) ||
+	if ((vma->vm_flags & (VM_DONTEXPAND | VM_RESERVED | VM_HUGETLB)) ||
 			vma == get_gate_vma(current)) {
 		if (lock)
 			make_pages_present(start, end);
@@ -443,7 +441,7 @@ static int do_mlock(unsigned long start,
 		prev = vma;
 
 	for (nstart = start ; ; ) {
-		unsigned int newflags;
+		unsigned long long newflags;
 
 		/* Here we know that  vma->vm_start <= nstart < vma->vm_end. */
 
@@ -524,7 +522,7 @@ static int do_mlockall(int flags)
 		goto out;
 
 	for (vma = current->mm->mmap; vma ; vma = prev->vm_next) {
-		unsigned int newflags;
+		unsigned long long newflags;
 
 		newflags = vma->vm_flags | VM_LOCKED;
 		if (!(flags & MCL_CURRENT))
--- 2.6.32-rc1/mm/mmap.c	2009-09-28 00:28:41.000000000 +0100
+++ ull_vm_flags/mm/mmap.c	2009-09-29 16:48:15.000000000 +0100
@@ -38,7 +38,7 @@
 #include "internal.h"
 
 #ifndef arch_mmap_check
-#define arch_mmap_check(addr, len, flags)	(0)
+#define arch_mmap_check(addr, len)			(0)
 #endif
 
 #ifndef arch_rebalance_pgtables
@@ -75,7 +75,7 @@ pgprot_t protection_map[16] = {
 	__S000, __S001, __S010, __S011, __S100, __S101, __S110, __S111
 };
 
-pgprot_t vm_get_page_prot(unsigned long vm_flags)
+pgprot_t vm_get_page_prot(unsigned long long vm_flags)
 {
 	return __pgprot(pgprot_val(protection_map[vm_flags &
 				(VM_READ|VM_WRITE|VM_EXEC|VM_SHARED)]) |
@@ -661,7 +661,7 @@ again:			remove_next = 1 + (end > next->
  * per-vma resources, so we don't attempt to merge those.
  */
 static inline int is_mergeable_vma(struct vm_area_struct *vma,
-			struct file *file, unsigned long vm_flags)
+			struct file *file, unsigned long long vm_flags)
 {
 	/* VM_CAN_NONLINEAR may get set later by f_op->mmap() */
 	if ((vma->vm_flags ^ vm_flags) & ~VM_CAN_NONLINEAR)
@@ -691,7 +691,7 @@ static inline int is_mergeable_anon_vma(
  * wrap, nor mmaps which cover the final page at index -1UL.
  */
 static int
-can_vma_merge_before(struct vm_area_struct *vma, unsigned long vm_flags,
+can_vma_merge_before(struct vm_area_struct *vma, unsigned long long vm_flags,
 	struct anon_vma *anon_vma, struct file *file, pgoff_t vm_pgoff)
 {
 	if (is_mergeable_vma(vma, file, vm_flags) &&
@@ -710,7 +710,7 @@ can_vma_merge_before(struct vm_area_stru
  * anon_vmas, nor if same anon_vma is assigned but offsets incompatible.
  */
 static int
-can_vma_merge_after(struct vm_area_struct *vma, unsigned long vm_flags,
+can_vma_merge_after(struct vm_area_struct *vma, unsigned long long vm_flags,
 	struct anon_vma *anon_vma, struct file *file, pgoff_t vm_pgoff)
 {
 	if (is_mergeable_vma(vma, file, vm_flags) &&
@@ -754,7 +754,7 @@ can_vma_merge_after(struct vm_area_struc
  */
 struct vm_area_struct *vma_merge(struct mm_struct *mm,
 			struct vm_area_struct *prev, unsigned long addr,
-			unsigned long end, unsigned long vm_flags,
+			unsigned long end, unsigned long long vm_flags,
 		     	struct anon_vma *anon_vma, struct file *file,
 			pgoff_t pgoff, struct mempolicy *policy)
 {
@@ -831,7 +831,7 @@ struct vm_area_struct *vma_merge(struct
 struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *vma)
 {
 	struct vm_area_struct *near;
-	unsigned long vm_flags;
+	unsigned long long vm_flags;
 
 	near = vma->vm_next;
 	if (!near)
@@ -885,19 +885,19 @@ none:
 }
 
 #ifdef CONFIG_PROC_FS
-void vm_stat_account(struct mm_struct *mm, unsigned long flags,
+void vm_stat_account(struct mm_struct *mm, unsigned long long vm_flags,
 						struct file *file, long pages)
 {
-	const unsigned long stack_flags
+	const unsigned long long stack_flags
 		= VM_STACK_FLAGS & (VM_GROWSUP|VM_GROWSDOWN);
 
 	if (file) {
 		mm->shared_vm += pages;
-		if ((flags & (VM_EXEC|VM_WRITE)) == VM_EXEC)
+		if ((vm_flags & (VM_EXEC|VM_WRITE)) == VM_EXEC)
 			mm->exec_vm += pages;
-	} else if (flags & stack_flags)
+	} else if (vm_flags & stack_flags)
 		mm->stack_vm += pages;
-	if (flags & (VM_RESERVED|VM_IO))
+	if (vm_flags & (VM_RESERVED|VM_IO))
 		mm->reserved_vm += pages;
 }
 #endif /* CONFIG_PROC_FS */
@@ -912,7 +912,7 @@ unsigned long do_mmap_pgoff(struct file
 {
 	struct mm_struct * mm = current->mm;
 	struct inode *inode;
-	unsigned int vm_flags;
+	unsigned long long vm_flags;
 	int error;
 	unsigned long reqprot = prot;
 
@@ -932,7 +932,7 @@ unsigned long do_mmap_pgoff(struct file
 	if (!(flags & MAP_FIXED))
 		addr = round_hint_to_min(addr);
 
-	error = arch_mmap_check(addr, len, flags);
+	error = arch_mmap_check(addr, len);
 	if (error)
 		return error;
 
@@ -1077,7 +1077,7 @@ EXPORT_SYMBOL(do_mmap_pgoff);
  */
 int vma_wants_writenotify(struct vm_area_struct *vma)
 {
-	unsigned int vm_flags = vma->vm_flags;
+	unsigned long long vm_flags = vma->vm_flags;
 
 	/* If it was private or non-writable, the write bit is already clear */
 	if ((vm_flags & (VM_WRITE|VM_SHARED)) != ((VM_WRITE|VM_SHARED)))
@@ -1105,7 +1105,8 @@ int vma_wants_writenotify(struct vm_area
  * We account for memory if it's a private writeable mapping,
  * not hugepages and VM_NORESERVE wasn't set.
  */
-static inline int accountable_mapping(struct file *file, unsigned int vm_flags)
+static inline int accountable_mapping(struct file *file,
+				      unsigned long long vm_flags)
 {
 	/*
 	 * hugetlb has its own accounting separate from the core VM
@@ -1119,7 +1120,7 @@ static inline int accountable_mapping(st
 
 unsigned long mmap_region(struct file *file, unsigned long addr,
 			  unsigned long len, unsigned long flags,
-			  unsigned int vm_flags, unsigned long pgoff)
+			  unsigned long long vm_flags, unsigned long pgoff)
 {
 	struct mm_struct *mm = current->mm;
 	struct vm_area_struct *vma, *prev;
@@ -1994,7 +1995,7 @@ unsigned long do_brk(unsigned long addr,
 {
 	struct mm_struct * mm = current->mm;
 	struct vm_area_struct * vma, * prev;
-	unsigned long flags;
+	unsigned long long vm_flags;
 	struct rb_node ** rb_link, * rb_parent;
 	pgoff_t pgoff = addr >> PAGE_SHIFT;
 	int error;
@@ -2013,9 +2014,7 @@ unsigned long do_brk(unsigned long addr,
 	if (error)
 		return error;
 
-	flags = VM_DATA_DEFAULT_FLAGS | VM_ACCOUNT | mm->def_flags;
-
-	error = arch_mmap_check(addr, len, flags);
+	error = arch_mmap_check(addr, len);
 	if (error)
 		return error;
 
@@ -2059,8 +2058,10 @@ unsigned long do_brk(unsigned long addr,
 	if (security_vm_enough_memory(len >> PAGE_SHIFT))
 		return -ENOMEM;
 
+	vm_flags = VM_DATA_DEFAULT_FLAGS | VM_ACCOUNT | mm->def_flags;
+
 	/* Can we just expand an old private anonymous mapping? */
-	vma = vma_merge(mm, prev, addr, addr + len, flags,
+	vma = vma_merge(mm, prev, addr, addr + len, vm_flags,
 					NULL, NULL, pgoff, NULL);
 	if (vma)
 		goto out;
@@ -2078,12 +2079,12 @@ unsigned long do_brk(unsigned long addr,
 	vma->vm_start = addr;
 	vma->vm_end = addr + len;
 	vma->vm_pgoff = pgoff;
-	vma->vm_flags = flags;
-	vma->vm_page_prot = vm_get_page_prot(flags);
+	vma->vm_flags = vm_flags;
+	vma->vm_page_prot = vm_get_page_prot(vm_flags);
 	vma_link(mm, vma, prev, rb_link, rb_parent);
 out:
 	mm->total_vm += len >> PAGE_SHIFT;
-	if (flags & VM_LOCKED) {
+	if (vm_flags & VM_LOCKED) {
 		if (!mlock_vma_pages_range(vma, addr, addr + len))
 			mm->locked_vm += (len >> PAGE_SHIFT);
 	}
@@ -2298,7 +2299,7 @@ static const struct vm_operations_struct
  */
 int install_special_mapping(struct mm_struct *mm,
 			    unsigned long addr, unsigned long len,
-			    unsigned long vm_flags, struct page **pages)
+			    unsigned long long vm_flags, struct page **pages)
 {
 	struct vm_area_struct *vma;
 
--- 2.6.32-rc1/mm/mprotect.c	2009-09-28 00:28:41.000000000 +0100
+++ ull_vm_flags/mm/mprotect.c	2009-09-29 16:48:15.000000000 +0100
@@ -134,10 +134,10 @@ static void change_protection(struct vm_
 
 int
 mprotect_fixup(struct vm_area_struct *vma, struct vm_area_struct **pprev,
-	unsigned long start, unsigned long end, unsigned long newflags)
+	unsigned long start, unsigned long end, unsigned long long newflags)
 {
 	struct mm_struct *mm = vma->vm_mm;
-	unsigned long oldflags = vma->vm_flags;
+	unsigned long long oldflags = vma->vm_flags;
 	long nrpages = (end - start) >> PAGE_SHIFT;
 	unsigned long charged = 0;
 	pgoff_t pgoff;
@@ -222,7 +222,7 @@ fail:
 SYSCALL_DEFINE3(mprotect, unsigned long, start, size_t, len,
 		unsigned long, prot)
 {
-	unsigned long vm_flags, nstart, end, tmp, reqprot;
+	unsigned long vm_prots, nstart, end, tmp, reqprot;
 	struct vm_area_struct *vma, *prev;
 	int error = -EINVAL;
 	const int grows = prot & (PROT_GROWSDOWN|PROT_GROWSUP);
@@ -248,7 +248,7 @@ SYSCALL_DEFINE3(mprotect, unsigned long,
 	if ((prot & PROT_READ) && (current->personality & READ_IMPLIES_EXEC))
 		prot |= PROT_EXEC;
 
-	vm_flags = calc_vm_prot_bits(prot);
+	vm_prots = calc_vm_prot_bits(prot);
 
 	down_write(&current->mm->mmap_sem);
 
@@ -278,11 +278,11 @@ SYSCALL_DEFINE3(mprotect, unsigned long,
 		prev = vma;
 
 	for (nstart = start ; ; ) {
-		unsigned long newflags;
+		unsigned long long newflags;
 
 		/* Here we know that  vma->vm_start <= nstart < vma->vm_end. */
 
-		newflags = vm_flags | (vma->vm_flags & ~(VM_READ | VM_WRITE | VM_EXEC));
+		newflags = vm_prots | (vma->vm_flags & ~(VM_READ | VM_WRITE | VM_EXEC));
 
 		/* newflags >> 4 shift VM_MAY% in place of VM_% */
 		if ((newflags & ~(newflags >> 4)) & (VM_READ | VM_WRITE | VM_EXEC)) {
--- 2.6.32-rc1/mm/mremap.c	2009-09-28 00:28:41.000000000 +0100
+++ ull_vm_flags/mm/mremap.c	2009-09-29 16:48:15.000000000 +0100
@@ -169,7 +169,7 @@ static unsigned long move_vma(struct vm_
 {
 	struct mm_struct *mm = vma->vm_mm;
 	struct vm_area_struct *new_vma;
-	unsigned long vm_flags = vma->vm_flags;
+	unsigned long long vm_flags = vma->vm_flags;
 	unsigned long new_pgoff;
 	unsigned long moved_len;
 	unsigned long excess = 0;
--- 2.6.32-rc1/mm/nommu.c	2009-09-28 00:28:41.000000000 +0100
+++ ull_vm_flags/mm/nommu.c	2009-09-29 16:48:15.000000000 +0100
@@ -134,7 +134,7 @@ int __get_user_pages(struct task_struct
 		     struct page **pages, struct vm_area_struct **vmas)
 {
 	struct vm_area_struct *vma;
-	unsigned long vm_flags;
+	unsigned long long vm_flags;
 	int i;
 
 	/* calculate required read or write permissions.
@@ -987,12 +987,12 @@ static int validate_mmap_request(struct
  * we've determined that we can make the mapping, now translate what we
  * now know into VMA flags
  */
-static unsigned long determine_vm_flags(struct file *file,
-					unsigned long prot,
-					unsigned long flags,
-					unsigned long capabilities)
+static unsigned long long determine_vm_flags(struct file *file,
+					     unsigned long prot,
+					     unsigned long flags,
+					     unsigned long capabilities)
 {
-	unsigned long vm_flags;
+	unsigned long long vm_flags;
 
 	vm_flags = calc_vm_prot_bits(prot) | calc_vm_flag_bits(flags);
 	vm_flags |= VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC;
@@ -1177,7 +1177,8 @@ unsigned long do_mmap_pgoff(struct file
 	struct vm_area_struct *vma;
 	struct vm_region *region;
 	struct rb_node *rb;
-	unsigned long capabilities, vm_flags, result;
+	unsigned long long vm_flags;
+	unsigned long capabilities, result;
 	int ret;
 
 	kenter(",%lx,%lx,%lx,%lx,%lx", addr, len, prot, flags, pgoff);
--- 2.6.32-rc1/mm/rmap.c	2009-09-28 00:28:41.000000000 +0100
+++ ull_vm_flags/mm/rmap.c	2009-09-29 16:48:15.000000000 +0100
@@ -340,7 +340,7 @@ int page_mapped_in_vma(struct page *page
 static int page_referenced_one(struct page *page,
 			       struct vm_area_struct *vma,
 			       unsigned int *mapcount,
-			       unsigned long *vm_flags)
+			       unsigned long long *vm_flags)
 {
 	struct mm_struct *mm = vma->vm_mm;
 	unsigned long address;
@@ -375,7 +375,7 @@ static int page_referenced_one(struct pa
 		 * mapping is already gone, the unmap path will have
 		 * set PG_referenced or activated the page.
 		 */
-		if (likely(!VM_SequentialReadHint(vma)))
+		if (likely(!(vma->vm_flags & VM_SEQ_READ)))
 			referenced++;
 	}
 
@@ -396,7 +396,7 @@ out:
 
 static int page_referenced_anon(struct page *page,
 				struct mem_cgroup *mem_cont,
-				unsigned long *vm_flags)
+				unsigned long long *vm_flags)
 {
 	unsigned int mapcount;
 	struct anon_vma *anon_vma;
@@ -441,7 +441,7 @@ static int page_referenced_anon(struct p
  */
 static int page_referenced_file(struct page *page,
 				struct mem_cgroup *mem_cont,
-				unsigned long *vm_flags)
+				unsigned long long *vm_flags)
 {
 	unsigned int mapcount;
 	struct address_space *mapping = page->mapping;
@@ -504,7 +504,7 @@ static int page_referenced_file(struct p
 int page_referenced(struct page *page,
 		    int is_locked,
 		    struct mem_cgroup *mem_cont,
-		    unsigned long *vm_flags)
+		    unsigned long long *vm_flags)
 {
 	int referenced = 0;
 
--- 2.6.32-rc1/mm/shmem.c	2009-09-28 00:28:41.000000000 +0100
+++ ull_vm_flags/mm/shmem.c	2009-09-29 16:48:15.000000000 +0100
@@ -2683,7 +2683,7 @@ int shmem_zero_setup(struct vm_area_stru
 	struct file *file;
 	loff_t size = vma->vm_end - vma->vm_start;
 
-	file = shmem_file_setup("dev/zero", size, vma->vm_flags);
+	file = shmem_file_setup("dev/zero", size, (unsigned long)vma->vm_flags);
 	if (IS_ERR(file))
 		return PTR_ERR(file);
 
--- 2.6.32-rc1/mm/vmscan.c	2009-09-28 00:28:41.000000000 +0100
+++ ull_vm_flags/mm/vmscan.c	2009-09-29 16:48:15.000000000 +0100
@@ -581,7 +581,7 @@ static unsigned long shrink_page_list(st
 	struct pagevec freed_pvec;
 	int pgactivate = 0;
 	unsigned long nr_reclaimed = 0;
-	unsigned long vm_flags;
+	unsigned long long vm_flags;
 
 	cond_resched();
 
@@ -1303,7 +1303,7 @@ static void shrink_active_list(unsigned
 {
 	unsigned long nr_taken;
 	unsigned long pgscanned;
-	unsigned long vm_flags;
+	unsigned long long vm_flags;
 	LIST_HEAD(l_hold);	/* The pages which were snipped off */
 	LIST_HEAD(l_active);
 	LIST_HEAD(l_inactive);

Re: No more bits in vm_area_struct's vm_flags.

[Christoph Lameter]

On Thu, 1 Oct 2009, Hugh Dickins wrote:

> Are we doing that?  If you have some example like, when PG_slab is set
> then PG_owner_priv_1 means such-and-such, but if not not: okay, I'm
> fine with that.

Look at how compound pages are handled in include/linux/page-flags.h

> But if you're saying something like, if PG_reclaim is set at the same
> time as PG_buddy, then they mean the page is not a buddy or under
> reclaim, but brokenbacked: then I'm a bit (or even 32 bits) worried.

Of course you need to be careful not to use two bits that can be used
indepedently.

> > VM_HUGETLB cannot grow up and down f.e. and there are
> > certainly lots of other impossible combinations that can be used to put
> > more information into the flags.
>
> Where it makes sense, where it's understandable, okay: there may be a
> few which could naturally use combinations.  But in general, no, I
> think we'd be asking for endless maintenance trouble if we change the
> meaning of some flags according to other flags.

We made the page flags stuff configurable. On 64 bit we use more flags, on
32 bit we compress the page flags a bit. Maybe do the same for vm_flags?

Re: No more bits in vm_area_struct's vm_flags.

[KAMEZAWA Hiroyuki]

On Thu, 1 Oct 2009 12:38:42 +0100 (BST)
Hugh Dickins <hugh.dickins@tiscali.co.uk> wrote:

> On Tue, 29 Sep 2009, KAMEZAWA Hiroyuki wrote:
> > 
> > (1) using "int"  will be never correct even on 32bit.
> > ==
> > vm_flags          242 arch/mips/mm/c-r3k.c 	int exec = vma->vm_flags & VM_EXEC;
> > vm_flags          293 drivers/char/mem.c 	return vma->vm_flags & VM_MAYSHARE;
> > vm_flags           44 mm/madvise.c   	int new_flags = vma->vm_flags;
> > vm_flags          547 mm/memory.c    	unsigned long vm_flags = vma->vm_flags;
> > 
> > But yes, it will be not a terrible bug for a while.
> 
> There may be a few, probably recently added and rarely used, vm_flags
> which we could consider moving from the low int to the high int of an
> unsigned long long, for cheaper operations where they're not involved.
> 
> But I've no intention to shuffle the majority of vm_flags around:
> VM_READ through VM_MAYSHARE will stay precisely where they are,
> and a lot of others beyond e.g. I don't see any need to make mm->
> def_flags an unsigned long long, so VM_LOCKED should stay low.
> 
> You say "using int will be never correct even on 32bit": you mean,
> when dealing with an unknown set of vm_flags, as in your madvise.c
> and memory.c examples above.  But the VM_EXEC example in arch/mips,
> I don't see any need to pester an arch maintainer to change it.
> The drivers/char/mem.c one I hadn't noticed (thanks), that one I'll
> happily change: not because it represents a bug, but just to set a
> good example to people adding such tests on higher flags in future.
> 

Sure.


> > 
> > (2) All vm macros should be defined with ULL suffix. for supporing ~ 
> > ==
> > vm_flags           30 arch/x86/mm/hugetlbpage.c 	unsigned long vm_flags = vma->vm_flags & ~VM_LOCKED;
> > 
> > (3) vma_merge()'s vm_flags should be ULL.
> 
> At first I thought you'd saved me a lot of embarrassment by mentioning
> those ULL suffixes, I hadn't put them in.  But after a quick test of
> what I thought was going to show a problem without them, no problem.
> Please would you send me a test program which demonstrates the need
> for all those ULLs?
> 
Ah, I'm sorry if I misunderstand C's rule. 

There are some places which use ~.
like
	vm_flags = vma->vm_flags & ~(VM_LOCKED);

~VM_LOCKED is 
	0xffffdfff or 0xffffffffffffdffff ?

Is my concern.

I tried following function on my old x86 box
==
#define FLAG    (0x20)

int foo(unsigned long long x)
{
        return x & ~FLAG;
}
==
(returning "int" as "bool")

compile this with gcc -S -O2 (gcc's version is 4.0)
==
foo:
        pushl   %ebp
        movl    %esp, %ebp
        movl    8(%ebp), %eax
        andl    $-33, %eax
        leave
        ret
==
Them, it seems higher bits are ignored for returning bool.


> Certainly the added higher flags would need ULLs to get through the
> compiler (or preprocessor); and I don't mind adding ULLs throughout
> as an example (though VM_EXEC will give ARM a slight problem), they
> do seem to shave about 30 bytes off my build, presumably nudge the
> compiler towards making better choices in a few places.  But I've
> not yet seen why they're necessary.
> 
> I am *not* intending to pursue this further right now: I'd still
> prefer us to be looking for vm_flags we can weed out.  But the day
> is sure to come when we need to extend, so here for the record (and
> for your curiosity!) is the patch I built up: complete so far as I
> know, except for a VMA_VM_FLAGS endian question in ARM (I'm pretty
> sure I'd get it wrong if I tried to adjust that one myself).
> 
thanks.

Regards,
-Kame

> arch_mmap_check() was amusing: I should probably extract that
> part and send it in as a separate patch, but no hurry.
> 
> I still don't understand why this adds 100 bytes to do_mmap_pgoff():
> things I experimented with (e.g. changing vm_stat_account back to
> taking unsigned long, changing the calc_vm algorithm back to not
> be mixing types) only made a few bytes difference.
> 
> Hugh
> 
>  arch/arm/include/asm/cacheflush.h            |    7 -
>  arch/arm/kernel/asm-offsets.c                |    3 
>  arch/ia64/include/asm/mman.h                 |    5 
>  arch/ia64/kernel/sys_ia64.c                  |    3 
>  arch/powerpc/include/asm/mman.h              |    6 
>  arch/s390/include/asm/mman.h                 |    2 
>  arch/sh/mm/tlbflush_64.c                     |    2 
>  arch/sparc/include/asm/mman.h                |    2 
>  arch/x86/mm/hugetlbpage.c                    |    4 
>  drivers/char/mem.c                           |    2 
>  drivers/infiniband/hw/ipath/ipath_file_ops.c |    4 
>  drivers/staging/android/binder.c             |    6 
>  fs/binfmt_elf_fdpic.c                        |   12 -
>  fs/exec.c                                    |    5 
>  include/linux/hugetlb.h                      |    2 
>  include/linux/ksm.h                          |    4 
>  include/linux/mm.h                           |  109 +++++++----------
>  include/linux/mm_types.h                     |    4 
>  include/linux/mman.h                         |    8 -
>  include/linux/rmap.h                         |    4 
>  mm/filemap.c                                 |    6 
>  mm/fremap.c                                  |    2 
>  mm/ksm.c                                     |    2 
>  mm/madvise.c                                 |    2 
>  mm/memory.c                                  |   12 -
>  mm/mlock.c                                   |   14 --
>  mm/mmap.c                                    |   53 ++++----
>  mm/mprotect.c                                |   12 -
>  mm/mremap.c                                  |    2 
>  mm/nommu.c                                   |   15 +-
>  mm/rmap.c                                    |   10 -
>  mm/shmem.c                                   |    2 
>  mm/vmscan.c                                  |    4 
>  33 files changed, 158 insertions(+), 172 deletions(-)
> 
> --- 2.6.32-rc1/arch/arm/include/asm/cacheflush.h	2009-09-28 00:27:55.000000000 +0100
> +++ ull_vm_flags/arch/arm/include/asm/cacheflush.h	2009-09-29 16:48:15.000000000 +0100
> @@ -169,7 +169,7 @@
>   *		specified address space before a change of page tables.
>   *		- start - user start address (inclusive, page aligned)
>   *		- end   - user end address   (exclusive, page aligned)
> - *		- flags - vma->vm_flags field
> + *		- flags - low unsigned long of vma->vm_flags field
>   *
>   *	coherent_kern_range(start, end)
>   *
> @@ -343,7 +343,7 @@ flush_cache_range(struct vm_area_struct
>  {
>  	if (cpumask_test_cpu(smp_processor_id(), mm_cpumask(vma->vm_mm)))
>  		__cpuc_flush_user_range(start & PAGE_MASK, PAGE_ALIGN(end),
> -					vma->vm_flags);
> +					(unsigned long)vma->vm_flags);
>  }
>  
>  static inline void
> @@ -351,7 +351,8 @@ flush_cache_page(struct vm_area_struct *
>  {
>  	if (cpumask_test_cpu(smp_processor_id(), mm_cpumask(vma->vm_mm))) {
>  		unsigned long addr = user_addr & PAGE_MASK;
> -		__cpuc_flush_user_range(addr, addr + PAGE_SIZE, vma->vm_flags);
> +		__cpuc_flush_user_range(addr, addr + PAGE_SIZE,
> +					(unsigned long)vma->vm_flags);
>  	}
>  }
>  
> --- 2.6.32-rc1/arch/arm/kernel/asm-offsets.c	2008-07-13 22:51:29.000000000 +0100
> +++ ull_vm_flags/arch/arm/kernel/asm-offsets.c	2009-09-29 16:48:15.000000000 +0100
> @@ -88,8 +88,9 @@ int main(void)
>  #endif
>    DEFINE(VMA_VM_MM,		offsetof(struct vm_area_struct, vm_mm));
>    DEFINE(VMA_VM_FLAGS,		offsetof(struct vm_area_struct, vm_flags));
> +				/* but that will be wrong for bigendian? */
>    BLANK();
> -  DEFINE(VM_EXEC,	       	VM_EXEC);
> +  DEFINE(VM_EXEC,	       	0x00000004);	/* mm.h now appends ULL */
>    BLANK();
>    DEFINE(PAGE_SZ,	       	PAGE_SIZE);
>    BLANK();
> --- 2.6.32-rc1/arch/ia64/include/asm/mman.h	2009-09-28 00:27:59.000000000 +0100
> +++ ull_vm_flags/arch/ia64/include/asm/mman.h	2009-09-29 16:48:15.000000000 +0100
> @@ -14,9 +14,8 @@
>  
>  #ifdef __KERNEL__
>  #ifndef __ASSEMBLY__
> -#define arch_mmap_check	ia64_mmap_check
> -int ia64_mmap_check(unsigned long addr, unsigned long len,
> -		unsigned long flags);
> +#define arch_mmap_check(addr, len)	ia64_mmap_check(addr, len)
> +int ia64_mmap_check(unsigned long addr, unsigned long len);
>  #endif
>  #endif
>  
> --- 2.6.32-rc1/arch/ia64/kernel/sys_ia64.c	2009-03-23 23:12:14.000000000 +0000
> +++ ull_vm_flags/arch/ia64/kernel/sys_ia64.c	2009-09-29 16:48:15.000000000 +0100
> @@ -169,8 +169,7 @@ sys_ia64_pipe (void)
>  	return retval;
>  }
>  
> -int ia64_mmap_check(unsigned long addr, unsigned long len,
> -		unsigned long flags)
> +int ia64_mmap_check(unsigned long addr, unsigned long len)
>  {
>  	unsigned long roff;
>  
> --- 2.6.32-rc1/arch/powerpc/include/asm/mman.h	2009-09-28 00:28:03.000000000 +0100
> +++ ull_vm_flags/arch/powerpc/include/asm/mman.h	2009-09-29 16:48:15.000000000 +0100
> @@ -38,13 +38,13 @@
>   * This file is included by linux/mman.h, so we can't use cacl_vm_prot_bits()
>   * here.  How important is the optimization?
>   */
> -static inline unsigned long arch_calc_vm_prot_bits(unsigned long prot)
> +static inline unsigned long long arch_calc_vm_prot_bits(unsigned long prot)
>  {
> -	return (prot & PROT_SAO) ? VM_SAO : 0;
> +	return (prot & PROT_SAO) ? VM_SAO : 0ULL;
>  }
>  #define arch_calc_vm_prot_bits(prot) arch_calc_vm_prot_bits(prot)
>  
> -static inline pgprot_t arch_vm_get_page_prot(unsigned long vm_flags)
> +static inline pgprot_t arch_vm_get_page_prot(unsigned long long vm_flags)
>  {
>  	return (vm_flags & VM_SAO) ? __pgprot(_PAGE_SAO) : __pgprot(0);
>  }
> --- 2.6.32-rc1/arch/s390/include/asm/mman.h	2009-09-28 00:28:04.000000000 +0100
> +++ ull_vm_flags/arch/s390/include/asm/mman.h	2009-09-29 16:48:15.000000000 +0100
> @@ -12,8 +12,8 @@
>  #include <asm-generic/mman.h>
>  
>  #if defined(__KERNEL__) && !defined(__ASSEMBLY__) && defined(CONFIG_64BIT)
> +#define arch_mmap_check(addr, len)	s390_mmap_check(addr, len)
>  int s390_mmap_check(unsigned long addr, unsigned long len);
> -#define arch_mmap_check(addr,len,flags)	s390_mmap_check(addr,len)
>  #endif
>  
>  #endif /* __S390_MMAN_H__ */
> --- 2.6.32-rc1/arch/sh/mm/tlbflush_64.c	2009-09-28 00:28:11.000000000 +0100
> +++ ull_vm_flags/arch/sh/mm/tlbflush_64.c	2009-09-29 16:48:15.000000000 +0100
> @@ -48,7 +48,7 @@ static inline void print_vma(struct vm_a
>  	printk("vma end   0x%08lx\n", vma->vm_end);
>  
>  	print_prots(vma->vm_page_prot);
> -	printk("vm_flags 0x%08lx\n", vma->vm_flags);
> +	printk("vm_flags 0x%08llx\n", vma->vm_flags);
>  }
>  
>  static inline void print_task(struct task_struct *tsk)
> --- 2.6.32-rc1/arch/sparc/include/asm/mman.h	2009-09-28 00:28:11.000000000 +0100
> +++ ull_vm_flags/arch/sparc/include/asm/mman.h	2009-09-29 16:48:15.000000000 +0100
> @@ -25,7 +25,7 @@
>  
>  #ifdef __KERNEL__
>  #ifndef __ASSEMBLY__
> -#define arch_mmap_check(addr,len,flags)	sparc_mmap_check(addr,len)
> +#define arch_mmap_check(addr, len)	sparc_mmap_check(addr, len)
>  int sparc_mmap_check(unsigned long addr, unsigned long len);
>  #endif
>  #endif
> --- 2.6.32-rc1/arch/x86/mm/hugetlbpage.c	2009-06-10 04:05:27.000000000 +0100
> +++ ull_vm_flags/arch/x86/mm/hugetlbpage.c	2009-09-29 16:48:15.000000000 +0100
> @@ -27,8 +27,8 @@ static unsigned long page_table_shareabl
>  	unsigned long s_end = sbase + PUD_SIZE;
>  
>  	/* Allow segments to share if only one is marked locked */
> -	unsigned long vm_flags = vma->vm_flags & ~VM_LOCKED;
> -	unsigned long svm_flags = svma->vm_flags & ~VM_LOCKED;
> +	unsigned long long vm_flags = vma->vm_flags & ~VM_LOCKED;
> +	unsigned long long svm_flags = svma->vm_flags & ~VM_LOCKED;
>  
>  	/*
>  	 * match the virtual addresses, permission and the alignment of the
> --- 2.6.32-rc1/drivers/char/mem.c	2009-09-28 00:28:15.000000000 +0100
> +++ ull_vm_flags/drivers/char/mem.c	2009-09-29 16:48:15.000000000 +0100
> @@ -290,7 +290,7 @@ static unsigned long get_unmapped_area_m
>  /* can't do an in-place private mapping if there's no MMU */
>  static inline int private_mapping_ok(struct vm_area_struct *vma)
>  {
> -	return vma->vm_flags & VM_MAYSHARE;
> +	return (vma->vm_flags & VM_MAYSHARE) != 0;
>  }
>  #else
>  #define get_unmapped_area_mem	NULL
> --- 2.6.32-rc1/drivers/infiniband/hw/ipath/ipath_file_ops.c	2009-09-28 00:28:17.000000000 +0100
> +++ ull_vm_flags/drivers/infiniband/hw/ipath/ipath_file_ops.c	2009-09-29 16:48:15.000000000 +0100
> @@ -1112,7 +1112,7 @@ static int mmap_rcvegrbufs(struct vm_are
>  
>  	if (vma->vm_flags & VM_WRITE) {
>  		dev_info(&dd->pcidev->dev, "Can't map eager buffers as "
> -			 "writable (flags=%lx)\n", vma->vm_flags);
> +			 "writable (flags=%llx)\n", vma->vm_flags);
>  		ret = -EPERM;
>  		goto bail;
>  	}
> @@ -1201,7 +1201,7 @@ static int mmap_kvaddr(struct vm_area_st
>                  if (vma->vm_flags & VM_WRITE) {
>                          dev_info(&dd->pcidev->dev,
>                                   "Can't map eager buffers as "
> -                                 "writable (flags=%lx)\n", vma->vm_flags);
> +                                 "writable (flags=%llx)\n", vma->vm_flags);
>                          ret = -EPERM;
>                          goto bail;
>                  }
> --- 2.6.32-rc1/drivers/staging/android/binder.c	2009-09-28 00:28:28.000000000 +0100
> +++ ull_vm_flags/drivers/staging/android/binder.c	2009-09-29 16:48:15.000000000 +0100
> @@ -2737,7 +2737,7 @@ static void binder_vma_open(struct vm_ar
>  {
>  	struct binder_proc *proc = vma->vm_private_data;
>  	binder_debug(BINDER_DEBUG_OPEN_CLOSE,
> -		     "binder: %d open vm area %lx-%lx (%ld K) vma %lx pagep %lx\n",
> +		     "binder: %d open vm area %lx-%lx (%ld K) vma %llx pagep %lx\n",
>  		     proc->pid, vma->vm_start, vma->vm_end,
>  		     (vma->vm_end - vma->vm_start) / SZ_1K, vma->vm_flags,
>  		     (unsigned long)pgprot_val(vma->vm_page_prot));
> @@ -2748,7 +2748,7 @@ static void binder_vma_close(struct vm_a
>  {
>  	struct binder_proc *proc = vma->vm_private_data;
>  	binder_debug(BINDER_DEBUG_OPEN_CLOSE,
> -		     "binder: %d close vm area %lx-%lx (%ld K) vma %lx pagep %lx\n",
> +		     "binder: %d close vm area %lx-%lx (%ld K) vma %llx pagep %lx\n",
>  		     proc->pid, vma->vm_start, vma->vm_end,
>  		     (vma->vm_end - vma->vm_start) / SZ_1K, vma->vm_flags,
>  		     (unsigned long)pgprot_val(vma->vm_page_prot));
> @@ -2773,7 +2773,7 @@ static int binder_mmap(struct file *filp
>  		vma->vm_end = vma->vm_start + SZ_4M;
>  
>  	binder_debug(BINDER_DEBUG_OPEN_CLOSE,
> -		     "binder_mmap: %d %lx-%lx (%ld K) vma %lx pagep %lx\n",
> +		     "binder_mmap: %d %lx-%lx (%ld K) vma %llx pagep %lx\n",
>  		     proc->pid, vma->vm_start, vma->vm_end,
>  		     (vma->vm_end - vma->vm_start) / SZ_1K, vma->vm_flags,
>  		     (unsigned long)pgprot_val(vma->vm_page_prot));
> --- 2.6.32-rc1/fs/binfmt_elf_fdpic.c	2009-09-28 00:28:34.000000000 +0100
> +++ ull_vm_flags/fs/binfmt_elf_fdpic.c	2009-09-29 16:48:15.000000000 +0100
> @@ -1235,7 +1235,7 @@ static int maydump(struct vm_area_struct
>  
>  	/* Do not dump I/O mapped devices or special mappings */
>  	if (vma->vm_flags & (VM_IO | VM_RESERVED)) {
> -		kdcore("%08lx: %08lx: no (IO)", vma->vm_start, vma->vm_flags);
> +		kdcore("%08lx: %08llx: no (IO)", vma->vm_start, vma->vm_flags);
>  		return 0;
>  	}
>  
> @@ -1243,7 +1243,7 @@ static int maydump(struct vm_area_struct
>  	 * them either. "dump_write()" can't handle it anyway.
>  	 */
>  	if (!(vma->vm_flags & VM_READ)) {
> -		kdcore("%08lx: %08lx: no (!read)", vma->vm_start, vma->vm_flags);
> +		kdcore("%08lx: %08llx: no (!read)", vma->vm_start, vma->vm_flags);
>  		return 0;
>  	}
>  
> @@ -1251,13 +1251,13 @@ static int maydump(struct vm_area_struct
>  	if (vma->vm_flags & VM_SHARED) {
>  		if (vma->vm_file->f_path.dentry->d_inode->i_nlink == 0) {
>  			dump_ok = test_bit(MMF_DUMP_ANON_SHARED, &mm_flags);
> -			kdcore("%08lx: %08lx: %s (share)", vma->vm_start,
> +			kdcore("%08lx: %08llx: %s (share)", vma->vm_start,
>  			       vma->vm_flags, dump_ok ? "yes" : "no");
>  			return dump_ok;
>  		}
>  
>  		dump_ok = test_bit(MMF_DUMP_MAPPED_SHARED, &mm_flags);
> -		kdcore("%08lx: %08lx: %s (share)", vma->vm_start,
> +		kdcore("%08lx: %08llx: %s (share)", vma->vm_start,
>  		       vma->vm_flags, dump_ok ? "yes" : "no");
>  		return dump_ok;
>  	}
> @@ -1266,14 +1266,14 @@ static int maydump(struct vm_area_struct
>  	/* By default, if it hasn't been written to, don't write it out */
>  	if (!vma->anon_vma) {
>  		dump_ok = test_bit(MMF_DUMP_MAPPED_PRIVATE, &mm_flags);
> -		kdcore("%08lx: %08lx: %s (!anon)", vma->vm_start,
> +		kdcore("%08lx: %08llx: %s (!anon)", vma->vm_start,
>  		       vma->vm_flags, dump_ok ? "yes" : "no");
>  		return dump_ok;
>  	}
>  #endif
>  
>  	dump_ok = test_bit(MMF_DUMP_ANON_PRIVATE, &mm_flags);
> -	kdcore("%08lx: %08lx: %s", vma->vm_start, vma->vm_flags,
> +	kdcore("%08lx: %08llx: %s", vma->vm_start, vma->vm_flags,
>  	       dump_ok ? "yes" : "no");
>  	return dump_ok;
>  }
> --- 2.6.32-rc1/fs/exec.c	2009-09-28 00:28:35.000000000 +0100
> +++ ull_vm_flags/fs/exec.c	2009-09-29 16:48:15.000000000 +0100
> @@ -570,7 +570,7 @@ int setup_arg_pages(struct linux_binprm
>  	struct mm_struct *mm = current->mm;
>  	struct vm_area_struct *vma = bprm->vma;
>  	struct vm_area_struct *prev = NULL;
> -	unsigned long vm_flags;
> +	unsigned long long vm_flags;
>  	unsigned long stack_base;
>  
>  #ifdef CONFIG_STACK_GROWSUP
> @@ -615,8 +615,7 @@ int setup_arg_pages(struct linux_binprm
>  		vm_flags &= ~VM_EXEC;
>  	vm_flags |= mm->def_flags;
>  
> -	ret = mprotect_fixup(vma, &prev, vma->vm_start, vma->vm_end,
> -			vm_flags);
> +	ret = mprotect_fixup(vma, &prev, vma->vm_start, vma->vm_end, vm_flags);
>  	if (ret)
>  		goto out_unlock;
>  	BUG_ON(prev != vma);
> --- 2.6.32-rc1/include/linux/hugetlb.h	2009-09-28 00:28:38.000000000 +0100
> +++ ull_vm_flags/include/linux/hugetlb.h	2009-09-29 16:48:15.000000000 +0100
> @@ -16,7 +16,7 @@ int PageHuge(struct page *page);
>  
>  static inline int is_vm_hugetlb_page(struct vm_area_struct *vma)
>  {
> -	return vma->vm_flags & VM_HUGETLB;
> +	return (vma->vm_flags & VM_HUGETLB) != 0;
>  }
>  
>  void reset_vma_resv_huge_pages(struct vm_area_struct *vma);
> --- 2.6.32-rc1/include/linux/ksm.h	2009-09-28 00:28:38.000000000 +0100
> +++ ull_vm_flags/include/linux/ksm.h	2009-09-29 16:48:15.000000000 +0100
> @@ -14,7 +14,7 @@
>  
>  #ifdef CONFIG_KSM
>  int ksm_madvise(struct vm_area_struct *vma, unsigned long start,
> -		unsigned long end, int advice, unsigned long *vm_flags);
> +		unsigned long end, int advice, unsigned long long *vm_flags);
>  int __ksm_enter(struct mm_struct *mm);
>  void __ksm_exit(struct mm_struct *mm);
>  
> @@ -54,7 +54,7 @@ static inline void page_add_ksm_rmap(str
>  #else  /* !CONFIG_KSM */
>  
>  static inline int ksm_madvise(struct vm_area_struct *vma, unsigned long start,
> -		unsigned long end, int advice, unsigned long *vm_flags)
> +		unsigned long end, int advice, unsigned long long *vm_flags)
>  {
>  	return 0;
>  }
> --- 2.6.32-rc1/include/linux/mm.h	2009-09-28 00:28:38.000000000 +0100
> +++ ull_vm_flags/include/linux/mm.h	2009-09-29 16:48:15.000000000 +0100
> @@ -65,46 +65,49 @@ extern unsigned int kobjsize(const void
>  /*
>   * vm_flags in vm_area_struct, see mm_types.h.
>   */
> -#define VM_READ		0x00000001	/* currently active flags */
> -#define VM_WRITE	0x00000002
> -#define VM_EXEC		0x00000004
> -#define VM_SHARED	0x00000008
> +#define VM_READ		0x00000001ULL	/* currently active flags */
> +#define VM_WRITE	0x00000002ULL
> +#define VM_EXEC		0x00000004ULL
> +#define VM_SHARED	0x00000008ULL
>  
>  /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
> -#define VM_MAYREAD	0x00000010	/* limits for mprotect() etc */
> -#define VM_MAYWRITE	0x00000020
> -#define VM_MAYEXEC	0x00000040
> -#define VM_MAYSHARE	0x00000080
> -
> -#define VM_GROWSDOWN	0x00000100	/* general info on the segment */
> -#define VM_GROWSUP	0x00000200
> -#define VM_PFNMAP	0x00000400	/* Page-ranges managed without "struct page", just pure PFN */
> -#define VM_DENYWRITE	0x00000800	/* ETXTBSY on write attempts.. */
> -
> -#define VM_EXECUTABLE	0x00001000
> -#define VM_LOCKED	0x00002000
> -#define VM_IO           0x00004000	/* Memory mapped I/O or similar */
> +#define VM_MAYREAD	0x00000010ULL	/* limits for mprotect() etc */
> +#define VM_MAYWRITE	0x00000020ULL
> +#define VM_MAYEXEC	0x00000040ULL
> +#define VM_MAYSHARE	0x00000080ULL
> +
> +#define VM_GROWSDOWN	0x00000100ULL	/* general info on the segment */
> +#define VM_GROWSUP	0x00000200ULL
> +#define VM_PFNMAP	0x00000400ULL	/* Page-ranges managed without "struct page", just pure PFN */
> +#define VM_DENYWRITE	0x00000800ULL	/* ETXTBSY on write attempts.. */
> +
> +#define VM_EXECUTABLE	0x00001000ULL
> +#define VM_LOCKED	0x00002000ULL
> +#define VM_IO   	0x00004000ULL	/* 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_SEQ_READ	0x00008000ULL	/* App will access data sequentially */
> +#define VM_RAND_READ	0x00010000ULL	/* 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 */
> -#define VM_ACCOUNT	0x00100000	/* Is a VM accounted object */
> -#define VM_NORESERVE	0x00200000	/* should the VM suppress accounting */
> -#define VM_HUGETLB	0x00400000	/* Huge TLB Page VM */
> -#define VM_NONLINEAR	0x00800000	/* Is non-linear (remap_file_pages) */
> -#define VM_MAPPED_COPY	0x01000000	/* T if mapped copy of data (nommu mmap) */
> -#define VM_INSERTPAGE	0x02000000	/* The vma has had "vm_insert_page()" done on it */
> -#define VM_ALWAYSDUMP	0x04000000	/* Always include in core dumps */
> -
> -#define VM_CAN_NONLINEAR 0x08000000	/* Has ->fault & does nonlinear pages */
> -#define VM_MIXEDMAP	0x10000000	/* Can contain "struct page" and pure PFN pages */
> -#define VM_SAO		0x20000000	/* Strong Access Ordering (powerpc) */
> -#define VM_PFN_AT_MMAP	0x40000000	/* PFNMAP vma that is fully mapped at mmap time */
> -#define VM_MERGEABLE	0x80000000	/* KSM may merge identical pages */
> +#define VM_DONTCOPY	0x00020000ULL	/* Do not copy this vma on fork */
> +#define VM_DONTEXPAND	0x00040000ULL	/* Cannot expand with mremap() */
> +#define VM_RESERVED	0x00080000ULL	/* Count as reserved_vm like IO */
> +
> +#define VM_ACCOUNT	0x00100000ULL	/* Is a VM accounted object */
> +#define VM_NORESERVE	0x00200000ULL	/* should the VM suppress accounting */
> +#define VM_HUGETLB	0x00400000ULL	/* Huge TLB Page VM */
> +#define VM_NONLINEAR	0x00800000ULL	/* Is non-linear (remap_file_pages) */
> +
> +#define VM_MAPPED_COPY	0x01000000ULL	/* T if mapped copy of data (nommu mmap) */
> +#define VM_INSERTPAGE	0x02000000ULL	/* The vma has had "vm_insert_page()" done on it */
> +#define VM_ALWAYSDUMP	0x04000000ULL	/* Always include in core dumps */
> +
> +#define VM_CAN_NONLINEAR 0x08000000ULL	/* Has ->fault & does nonlinear pages */
> +
> +#define VM_MIXEDMAP	0x10000000ULL	/* Can contain "struct page" and pure PFN pages */
> +#define VM_SAO		0x20000000ULL	/* Strong Access Ordering (powerpc) */
> +#define VM_PFN_AT_MMAP	0x40000000ULL	/* PFNMAP vma that is fully mapped at mmap time */
> +#define VM_MERGEABLE	0x80000000ULL	/* KSM may merge identical pages */
>  
>  #ifndef VM_STACK_DEFAULT_FLAGS		/* arch can override this */
>  #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
> @@ -116,12 +119,6 @@ extern unsigned int kobjsize(const void
>  #define VM_STACK_FLAGS	(VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
>  #endif
>  
> -#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)
> -
>  /*
>   * special vmas that are non-mergable, non-mlock()able
>   */
> @@ -147,12 +144,12 @@ extern pgprot_t protection_map[16];
>   */
>  static inline int is_linear_pfn_mapping(struct vm_area_struct *vma)
>  {
> -	return (vma->vm_flags & VM_PFN_AT_MMAP);
> +	return (vma->vm_flags & VM_PFN_AT_MMAP) != 0;
>  }
>  
>  static inline int is_pfn_mapping(struct vm_area_struct *vma)
>  {
> -	return (vma->vm_flags & VM_PFNMAP);
> +	return (vma->vm_flags & VM_PFNMAP) != 0;
>  }
>  
>  /*
> @@ -715,14 +712,6 @@ int shmem_lock(struct file *file, int lo
>  struct file *shmem_file_setup(const char *name, loff_t size, unsigned long flags);
>  int shmem_zero_setup(struct vm_area_struct *);
>  
> -#ifndef CONFIG_MMU
> -extern unsigned long shmem_get_unmapped_area(struct file *file,
> -					     unsigned long addr,
> -					     unsigned long len,
> -					     unsigned long pgoff,
> -					     unsigned long flags);
> -#endif
> -
>  extern int can_do_mlock(void);
>  extern int user_shm_lock(size_t, struct user_struct *);
>  extern void user_shm_unlock(size_t, struct user_struct *);
> @@ -845,7 +834,7 @@ extern unsigned long do_mremap(unsigned
>  			       unsigned long flags, unsigned long new_addr);
>  extern int mprotect_fixup(struct vm_area_struct *vma,
>  			  struct vm_area_struct **pprev, unsigned long start,
> -			  unsigned long end, unsigned long newflags);
> +			  unsigned long end, unsigned long long newflags);
>  
>  /*
>   * doesn't attempt to fault and will return short.
> @@ -1096,7 +1085,7 @@ extern void vma_adjust(struct vm_area_st
>  	unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
>  extern struct vm_area_struct *vma_merge(struct mm_struct *,
>  	struct vm_area_struct *prev, unsigned long addr, unsigned long end,
> -	unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
> +	unsigned long long vm_flags, struct anon_vma *, struct file *, pgoff_t,
>  	struct mempolicy *);
>  extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
>  extern int split_vma(struct mm_struct *,
> @@ -1125,9 +1114,8 @@ static inline void removed_exe_file_vma(
>  #endif /* CONFIG_PROC_FS */
>  
>  extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
> -extern int install_special_mapping(struct mm_struct *mm,
> -				   unsigned long addr, unsigned long len,
> -				   unsigned long flags, struct page **pages);
> +extern int install_special_mapping(struct mm_struct *mm, unsigned long addr,
> +	unsigned long len, unsigned long long vm_flags, struct page **pages);
>  
>  extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
>  
> @@ -1136,7 +1124,7 @@ extern unsigned long do_mmap_pgoff(struc
>  	unsigned long flag, unsigned long pgoff);
>  extern unsigned long mmap_region(struct file *file, unsigned long addr,
>  	unsigned long len, unsigned long flags,
> -	unsigned int vm_flags, unsigned long pgoff);
> +	unsigned long long vm_flags, unsigned long pgoff);
>  
>  static inline unsigned long do_mmap(struct file *file, unsigned long addr,
>  	unsigned long len, unsigned long prot,
> @@ -1222,7 +1210,7 @@ static inline unsigned long vma_pages(st
>  	return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
>  }
>  
> -pgprot_t vm_get_page_prot(unsigned long vm_flags);
> +pgprot_t vm_get_page_prot(unsigned long long vm_flags);
>  struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
>  int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
>  			unsigned long pfn, unsigned long size, pgprot_t);
> @@ -1246,10 +1234,11 @@ extern int apply_to_page_range(struct mm
>  			       unsigned long size, pte_fn_t fn, void *data);
>  
>  #ifdef CONFIG_PROC_FS
> -void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
> +void vm_stat_account(struct mm_struct *mm,
> +		unsigned long long vm_flags, struct file *file, long pages);
>  #else
>  static inline void vm_stat_account(struct mm_struct *mm,
> -			unsigned long flags, struct file *file, long pages)
> +		unsigned long long vm_flags, struct file *file, long pages)
>  {
>  }
>  #endif /* CONFIG_PROC_FS */
> --- 2.6.32-rc1/include/linux/mm_types.h	2009-09-28 00:28:38.000000000 +0100
> +++ ull_vm_flags/include/linux/mm_types.h	2009-09-29 16:48:15.000000000 +0100
> @@ -115,7 +115,7 @@ struct page {
>   */
>  struct vm_region {
>  	struct rb_node	vm_rb;		/* link in global region tree */
> -	unsigned long	vm_flags;	/* VMA vm_flags */
> +	unsigned long long vm_flags;	/* VMA vm_flags */
>  	unsigned long	vm_start;	/* start address of region */
>  	unsigned long	vm_end;		/* region initialised to here */
>  	unsigned long	vm_top;		/* region allocated to here */
> @@ -141,7 +141,7 @@ struct vm_area_struct {
>  	struct vm_area_struct *vm_next;
>  
>  	pgprot_t vm_page_prot;		/* Access permissions of this VMA. */
> -	unsigned long vm_flags;		/* Flags, see mm.h. */
> +	unsigned long long vm_flags;	/* Flags, see mm.h. */
>  
>  	struct rb_node vm_rb;
>  
> --- 2.6.32-rc1/include/linux/mman.h	2009-06-10 04:05:27.000000000 +0100
> +++ ull_vm_flags/include/linux/mman.h	2009-09-29 16:48:15.000000000 +0100
> @@ -35,7 +35,7 @@ static inline void vm_unacct_memory(long
>   */
>  
>  #ifndef arch_calc_vm_prot_bits
> -#define arch_calc_vm_prot_bits(prot) 0
> +#define arch_calc_vm_prot_bits(prot)	0ULL
>  #endif
>  
>  #ifndef arch_vm_get_page_prot
> @@ -69,8 +69,7 @@ static inline int arch_validate_prot(uns
>  /*
>   * Combine the mmap "prot" argument into "vm_flags" used internally.
>   */
> -static inline unsigned long
> -calc_vm_prot_bits(unsigned long prot)
> +static inline unsigned long long calc_vm_prot_bits(unsigned long prot)
>  {
>  	return _calc_vm_trans(prot, PROT_READ,  VM_READ ) |
>  	       _calc_vm_trans(prot, PROT_WRITE, VM_WRITE) |
> @@ -81,8 +80,7 @@ calc_vm_prot_bits(unsigned long prot)
>  /*
>   * Combine the mmap "flags" argument into "vm_flags" used internally.
>   */
> -static inline unsigned long
> -calc_vm_flag_bits(unsigned long flags)
> +static inline unsigned long long calc_vm_flag_bits(unsigned long flags)
>  {
>  	return _calc_vm_trans(flags, MAP_GROWSDOWN,  VM_GROWSDOWN ) |
>  	       _calc_vm_trans(flags, MAP_DENYWRITE,  VM_DENYWRITE ) |
> --- 2.6.32-rc1/include/linux/rmap.h	2009-09-28 00:28:39.000000000 +0100
> +++ ull_vm_flags/include/linux/rmap.h	2009-09-29 16:48:15.000000000 +0100
> @@ -80,7 +80,7 @@ static inline void page_dup_rmap(struct
>   * Called from mm/vmscan.c to handle paging out
>   */
>  int page_referenced(struct page *, int is_locked,
> -			struct mem_cgroup *cnt, unsigned long *vm_flags);
> +			struct mem_cgroup *cnt, unsigned long long *vm_flags);
>  enum ttu_flags {
>  	TTU_UNMAP = 0,			/* unmap mode */
>  	TTU_MIGRATION = 1,		/* migration mode */
> @@ -135,7 +135,7 @@ int page_mapped_in_vma(struct page *page
>  
>  static inline int page_referenced(struct page *page, int is_locked,
>  				  struct mem_cgroup *cnt,
> -				  unsigned long *vm_flags)
> +				  unsigned long long *vm_flags)
>  {
>  	*vm_flags = 0;
>  	return TestClearPageReferenced(page);
> --- 2.6.32-rc1/mm/filemap.c	2009-09-28 00:28:41.000000000 +0100
> +++ ull_vm_flags/mm/filemap.c	2009-09-29 16:48:15.000000000 +0100
> @@ -1429,10 +1429,10 @@ static void do_sync_mmap_readahead(struc
>  	struct address_space *mapping = file->f_mapping;
>  
>  	/* If we don't want any read-ahead, don't bother */
> -	if (VM_RandomReadHint(vma))
> +	if (vma->vm_flags & VM_RAND_READ)
>  		return;
>  
> -	if (VM_SequentialReadHint(vma) ||
> +	if ((vma->vm_flags & VM_SEQ_READ) ||
>  			offset - 1 == (ra->prev_pos >> PAGE_CACHE_SHIFT)) {
>  		page_cache_sync_readahead(mapping, ra, file, offset,
>  					  ra->ra_pages);
> @@ -1474,7 +1474,7 @@ static void do_async_mmap_readahead(stru
>  	struct address_space *mapping = file->f_mapping;
>  
>  	/* If we don't want any read-ahead, don't bother */
> -	if (VM_RandomReadHint(vma))
> +	if (vma->vm_flags & VM_RAND_READ)
>  		return;
>  	if (ra->mmap_miss > 0)
>  		ra->mmap_miss--;
> --- 2.6.32-rc1/mm/fremap.c	2009-03-23 23:12:14.000000000 +0000
> +++ ull_vm_flags/mm/fremap.c	2009-09-29 16:48:15.000000000 +0100
> @@ -221,7 +221,7 @@ SYSCALL_DEFINE5(remap_file_pages, unsign
>  		/*
>  		 * drop PG_Mlocked flag for over-mapped range
>  		 */
> -		unsigned int saved_flags = vma->vm_flags;
> +		unsigned long long saved_flags = vma->vm_flags;
>  		munlock_vma_pages_range(vma, start, start + size);
>  		vma->vm_flags = saved_flags;
>  	}
> --- 2.6.32-rc1/mm/ksm.c	2009-09-28 00:28:41.000000000 +0100
> +++ ull_vm_flags/mm/ksm.c	2009-09-29 16:48:15.000000000 +0100
> @@ -1366,7 +1366,7 @@ static int ksm_scan_thread(void *nothing
>  }
>  
>  int ksm_madvise(struct vm_area_struct *vma, unsigned long start,
> -		unsigned long end, int advice, unsigned long *vm_flags)
> +		unsigned long end, int advice, unsigned long long *vm_flags)
>  {
>  	struct mm_struct *mm = vma->vm_mm;
>  	int err;
> --- 2.6.32-rc1/mm/madvise.c	2009-09-28 00:28:41.000000000 +0100
> +++ ull_vm_flags/mm/madvise.c	2009-09-29 16:48:15.000000000 +0100
> @@ -42,7 +42,7 @@ static long madvise_behavior(struct vm_a
>  	struct mm_struct * mm = vma->vm_mm;
>  	int error = 0;
>  	pgoff_t pgoff;
> -	unsigned long new_flags = vma->vm_flags;
> +	unsigned long long new_flags = vma->vm_flags;
>  
>  	switch (behavior) {
>  	case MADV_NORMAL:
> --- 2.6.32-rc1/mm/memory.c	2009-09-28 00:28:41.000000000 +0100
> +++ ull_vm_flags/mm/memory.c	2009-09-29 16:48:15.000000000 +0100
> @@ -437,7 +437,7 @@ static void print_bad_pte(struct vm_area
>  		page_mapcount(page), page->mapping, page->index);
>  	}
>  	printk(KERN_ALERT
> -		"addr:%p vm_flags:%08lx anon_vma:%p mapping:%p index:%lx\n",
> +		"addr:%p vm_flags:%08llx anon_vma:%p mapping:%p index:%lx\n",
>  		(void *)addr, vma->vm_flags, vma->anon_vma, mapping, index);
>  	/*
>  	 * Choose text because data symbols depend on CONFIG_KALLSYMS_ALL=y
> @@ -452,9 +452,9 @@ static void print_bad_pte(struct vm_area
>  	add_taint(TAINT_BAD_PAGE);
>  }
>  
> -static inline int is_cow_mapping(unsigned int flags)
> +static inline int is_cow_mapping(unsigned long long vm_flags)
>  {
> -	return (flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE;
> +	return (vm_flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE;
>  }
>  
>  #ifndef is_zero_pfn
> @@ -577,7 +577,7 @@ copy_one_pte(struct mm_struct *dst_mm, s
>  		pte_t *dst_pte, pte_t *src_pte, struct vm_area_struct *vma,
>  		unsigned long addr, int *rss)
>  {
> -	unsigned long vm_flags = vma->vm_flags;
> +	unsigned long long vm_flags = vma->vm_flags;
>  	pte_t pte = *src_pte;
>  	struct page *page;
>  
> @@ -852,7 +852,7 @@ static unsigned long zap_pte_range(struc
>  				if (pte_dirty(ptent))
>  					set_page_dirty(page);
>  				if (pte_young(ptent) &&
> -				    likely(!VM_SequentialReadHint(vma)))
> +				    likely((!(vma->vm_flags & VM_SEQ_READ))))
>  					mark_page_accessed(page);
>  				file_rss--;
>  			}
> @@ -1231,7 +1231,7 @@ int __get_user_pages(struct task_struct
>  		     struct page **pages, struct vm_area_struct **vmas)
>  {
>  	int i;
> -	unsigned long vm_flags;
> +	unsigned long long vm_flags;
>  
>  	if (nr_pages <= 0)
>  		return 0;
> --- 2.6.32-rc1/mm/mlock.c	2009-09-28 00:28:41.000000000 +0100
> +++ ull_vm_flags/mm/mlock.c	2009-09-29 16:48:15.000000000 +0100
> @@ -266,8 +266,7 @@ long mlock_vma_pages_range(struct vm_are
>  	if (vma->vm_flags & (VM_IO | VM_PFNMAP))
>  		goto no_mlock;
>  
> -	if (!((vma->vm_flags & (VM_DONTEXPAND | VM_RESERVED)) ||
> -			is_vm_hugetlb_page(vma) ||
> +	if (!((vma->vm_flags & (VM_DONTEXPAND | VM_RESERVED | VM_HUGETLB)) ||
>  			vma == get_gate_vma(current))) {
>  
>  		__mlock_vma_pages_range(vma, start, end);
> @@ -354,20 +353,19 @@ void munlock_vma_pages_range(struct vm_a
>   * For vmas that pass the filters, merge/split as appropriate.
>   */
>  static int mlock_fixup(struct vm_area_struct *vma, struct vm_area_struct **prev,
> -	unsigned long start, unsigned long end, unsigned int newflags)
> +	unsigned long start, unsigned long end, unsigned long long newflags)
>  {
>  	struct mm_struct *mm = vma->vm_mm;
>  	pgoff_t pgoff;
>  	int nr_pages;
>  	int ret = 0;
> -	int lock = newflags & VM_LOCKED;
> +	int lock = !!(newflags & VM_LOCKED);
>  
>  	if (newflags == vma->vm_flags ||
>  			(vma->vm_flags & (VM_IO | VM_PFNMAP)))
>  		goto out;	/* don't set VM_LOCKED,  don't count */
>  
> -	if ((vma->vm_flags & (VM_DONTEXPAND | VM_RESERVED)) ||
> -			is_vm_hugetlb_page(vma) ||
> +	if ((vma->vm_flags & (VM_DONTEXPAND | VM_RESERVED | VM_HUGETLB)) ||
>  			vma == get_gate_vma(current)) {
>  		if (lock)
>  			make_pages_present(start, end);
> @@ -443,7 +441,7 @@ static int do_mlock(unsigned long start,
>  		prev = vma;
>  
>  	for (nstart = start ; ; ) {
> -		unsigned int newflags;
> +		unsigned long long newflags;
>  
>  		/* Here we know that  vma->vm_start <= nstart < vma->vm_end. */
>  
> @@ -524,7 +522,7 @@ static int do_mlockall(int flags)
>  		goto out;
>  
>  	for (vma = current->mm->mmap; vma ; vma = prev->vm_next) {
> -		unsigned int newflags;
> +		unsigned long long newflags;
>  
>  		newflags = vma->vm_flags | VM_LOCKED;
>  		if (!(flags & MCL_CURRENT))
> --- 2.6.32-rc1/mm/mmap.c	2009-09-28 00:28:41.000000000 +0100
> +++ ull_vm_flags/mm/mmap.c	2009-09-29 16:48:15.000000000 +0100
> @@ -38,7 +38,7 @@
>  #include "internal.h"
>  
>  #ifndef arch_mmap_check
> -#define arch_mmap_check(addr, len, flags)	(0)
> +#define arch_mmap_check(addr, len)			(0)
>  #endif
>  
>  #ifndef arch_rebalance_pgtables
> @@ -75,7 +75,7 @@ pgprot_t protection_map[16] = {
>  	__S000, __S001, __S010, __S011, __S100, __S101, __S110, __S111
>  };
>  
> -pgprot_t vm_get_page_prot(unsigned long vm_flags)
> +pgprot_t vm_get_page_prot(unsigned long long vm_flags)
>  {
>  	return __pgprot(pgprot_val(protection_map[vm_flags &
>  				(VM_READ|VM_WRITE|VM_EXEC|VM_SHARED)]) |
> @@ -661,7 +661,7 @@ again:			remove_next = 1 + (end > next->
>   * per-vma resources, so we don't attempt to merge those.
>   */
>  static inline int is_mergeable_vma(struct vm_area_struct *vma,
> -			struct file *file, unsigned long vm_flags)
> +			struct file *file, unsigned long long vm_flags)
>  {
>  	/* VM_CAN_NONLINEAR may get set later by f_op->mmap() */
>  	if ((vma->vm_flags ^ vm_flags) & ~VM_CAN_NONLINEAR)
> @@ -691,7 +691,7 @@ static inline int is_mergeable_anon_vma(
>   * wrap, nor mmaps which cover the final page at index -1UL.
>   */
>  static int
> -can_vma_merge_before(struct vm_area_struct *vma, unsigned long vm_flags,
> +can_vma_merge_before(struct vm_area_struct *vma, unsigned long long vm_flags,
>  	struct anon_vma *anon_vma, struct file *file, pgoff_t vm_pgoff)
>  {
>  	if (is_mergeable_vma(vma, file, vm_flags) &&
> @@ -710,7 +710,7 @@ can_vma_merge_before(struct vm_area_stru
>   * anon_vmas, nor if same anon_vma is assigned but offsets incompatible.
>   */
>  static int
> -can_vma_merge_after(struct vm_area_struct *vma, unsigned long vm_flags,
> +can_vma_merge_after(struct vm_area_struct *vma, unsigned long long vm_flags,
>  	struct anon_vma *anon_vma, struct file *file, pgoff_t vm_pgoff)
>  {
>  	if (is_mergeable_vma(vma, file, vm_flags) &&
> @@ -754,7 +754,7 @@ can_vma_merge_after(struct vm_area_struc
>   */
>  struct vm_area_struct *vma_merge(struct mm_struct *mm,
>  			struct vm_area_struct *prev, unsigned long addr,
> -			unsigned long end, unsigned long vm_flags,
> +			unsigned long end, unsigned long long vm_flags,
>  		     	struct anon_vma *anon_vma, struct file *file,
>  			pgoff_t pgoff, struct mempolicy *policy)
>  {
> @@ -831,7 +831,7 @@ struct vm_area_struct *vma_merge(struct
>  struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *vma)
>  {
>  	struct vm_area_struct *near;
> -	unsigned long vm_flags;
> +	unsigned long long vm_flags;
>  
>  	near = vma->vm_next;
>  	if (!near)
> @@ -885,19 +885,19 @@ none:
>  }
>  
>  #ifdef CONFIG_PROC_FS
> -void vm_stat_account(struct mm_struct *mm, unsigned long flags,
> +void vm_stat_account(struct mm_struct *mm, unsigned long long vm_flags,
>  						struct file *file, long pages)
>  {
> -	const unsigned long stack_flags
> +	const unsigned long long stack_flags
>  		= VM_STACK_FLAGS & (VM_GROWSUP|VM_GROWSDOWN);
>  
>  	if (file) {
>  		mm->shared_vm += pages;
> -		if ((flags & (VM_EXEC|VM_WRITE)) == VM_EXEC)
> +		if ((vm_flags & (VM_EXEC|VM_WRITE)) == VM_EXEC)
>  			mm->exec_vm += pages;
> -	} else if (flags & stack_flags)
> +	} else if (vm_flags & stack_flags)
>  		mm->stack_vm += pages;
> -	if (flags & (VM_RESERVED|VM_IO))
> +	if (vm_flags & (VM_RESERVED|VM_IO))
>  		mm->reserved_vm += pages;
>  }
>  #endif /* CONFIG_PROC_FS */
> @@ -912,7 +912,7 @@ unsigned long do_mmap_pgoff(struct file
>  {
>  	struct mm_struct * mm = current->mm;
>  	struct inode *inode;
> -	unsigned int vm_flags;
> +	unsigned long long vm_flags;
>  	int error;
>  	unsigned long reqprot = prot;
>  
> @@ -932,7 +932,7 @@ unsigned long do_mmap_pgoff(struct file
>  	if (!(flags & MAP_FIXED))
>  		addr = round_hint_to_min(addr);
>  
> -	error = arch_mmap_check(addr, len, flags);
> +	error = arch_mmap_check(addr, len);
>  	if (error)
>  		return error;
>  
> @@ -1077,7 +1077,7 @@ EXPORT_SYMBOL(do_mmap_pgoff);
>   */
>  int vma_wants_writenotify(struct vm_area_struct *vma)
>  {
> -	unsigned int vm_flags = vma->vm_flags;
> +	unsigned long long vm_flags = vma->vm_flags;
>  
>  	/* If it was private or non-writable, the write bit is already clear */
>  	if ((vm_flags & (VM_WRITE|VM_SHARED)) != ((VM_WRITE|VM_SHARED)))
> @@ -1105,7 +1105,8 @@ int vma_wants_writenotify(struct vm_area
>   * We account for memory if it's a private writeable mapping,
>   * not hugepages and VM_NORESERVE wasn't set.
>   */
> -static inline int accountable_mapping(struct file *file, unsigned int vm_flags)
> +static inline int accountable_mapping(struct file *file,
> +				      unsigned long long vm_flags)
>  {
>  	/*
>  	 * hugetlb has its own accounting separate from the core VM
> @@ -1119,7 +1120,7 @@ static inline int accountable_mapping(st
>  
>  unsigned long mmap_region(struct file *file, unsigned long addr,
>  			  unsigned long len, unsigned long flags,
> -			  unsigned int vm_flags, unsigned long pgoff)
> +			  unsigned long long vm_flags, unsigned long pgoff)
>  {
>  	struct mm_struct *mm = current->mm;
>  	struct vm_area_struct *vma, *prev;
> @@ -1994,7 +1995,7 @@ unsigned long do_brk(unsigned long addr,
>  {
>  	struct mm_struct * mm = current->mm;
>  	struct vm_area_struct * vma, * prev;
> -	unsigned long flags;
> +	unsigned long long vm_flags;
>  	struct rb_node ** rb_link, * rb_parent;
>  	pgoff_t pgoff = addr >> PAGE_SHIFT;
>  	int error;
> @@ -2013,9 +2014,7 @@ unsigned long do_brk(unsigned long addr,
>  	if (error)
>  		return error;
>  
> -	flags = VM_DATA_DEFAULT_FLAGS | VM_ACCOUNT | mm->def_flags;
> -
> -	error = arch_mmap_check(addr, len, flags);
> +	error = arch_mmap_check(addr, len);
>  	if (error)
>  		return error;
>  
> @@ -2059,8 +2058,10 @@ unsigned long do_brk(unsigned long addr,
>  	if (security_vm_enough_memory(len >> PAGE_SHIFT))
>  		return -ENOMEM;
>  
> +	vm_flags = VM_DATA_DEFAULT_FLAGS | VM_ACCOUNT | mm->def_flags;
> +
>  	/* Can we just expand an old private anonymous mapping? */
> -	vma = vma_merge(mm, prev, addr, addr + len, flags,
> +	vma = vma_merge(mm, prev, addr, addr + len, vm_flags,
>  					NULL, NULL, pgoff, NULL);
>  	if (vma)
>  		goto out;
> @@ -2078,12 +2079,12 @@ unsigned long do_brk(unsigned long addr,
>  	vma->vm_start = addr;
>  	vma->vm_end = addr + len;
>  	vma->vm_pgoff = pgoff;
> -	vma->vm_flags = flags;
> -	vma->vm_page_prot = vm_get_page_prot(flags);
> +	vma->vm_flags = vm_flags;
> +	vma->vm_page_prot = vm_get_page_prot(vm_flags);
>  	vma_link(mm, vma, prev, rb_link, rb_parent);
>  out:
>  	mm->total_vm += len >> PAGE_SHIFT;
> -	if (flags & VM_LOCKED) {
> +	if (vm_flags & VM_LOCKED) {
>  		if (!mlock_vma_pages_range(vma, addr, addr + len))
>  			mm->locked_vm += (len >> PAGE_SHIFT);
>  	}
> @@ -2298,7 +2299,7 @@ static const struct vm_operations_struct
>   */
>  int install_special_mapping(struct mm_struct *mm,
>  			    unsigned long addr, unsigned long len,
> -			    unsigned long vm_flags, struct page **pages)
> +			    unsigned long long vm_flags, struct page **pages)
>  {
>  	struct vm_area_struct *vma;
>  
> --- 2.6.32-rc1/mm/mprotect.c	2009-09-28 00:28:41.000000000 +0100
> +++ ull_vm_flags/mm/mprotect.c	2009-09-29 16:48:15.000000000 +0100
> @@ -134,10 +134,10 @@ static void change_protection(struct vm_
>  
>  int
>  mprotect_fixup(struct vm_area_struct *vma, struct vm_area_struct **pprev,
> -	unsigned long start, unsigned long end, unsigned long newflags)
> +	unsigned long start, unsigned long end, unsigned long long newflags)
>  {
>  	struct mm_struct *mm = vma->vm_mm;
> -	unsigned long oldflags = vma->vm_flags;
> +	unsigned long long oldflags = vma->vm_flags;
>  	long nrpages = (end - start) >> PAGE_SHIFT;
>  	unsigned long charged = 0;
>  	pgoff_t pgoff;
> @@ -222,7 +222,7 @@ fail:
>  SYSCALL_DEFINE3(mprotect, unsigned long, start, size_t, len,
>  		unsigned long, prot)
>  {
> -	unsigned long vm_flags, nstart, end, tmp, reqprot;
> +	unsigned long vm_prots, nstart, end, tmp, reqprot;
>  	struct vm_area_struct *vma, *prev;
>  	int error = -EINVAL;
>  	const int grows = prot & (PROT_GROWSDOWN|PROT_GROWSUP);
> @@ -248,7 +248,7 @@ SYSCALL_DEFINE3(mprotect, unsigned long,
>  	if ((prot & PROT_READ) && (current->personality & READ_IMPLIES_EXEC))
>  		prot |= PROT_EXEC;
>  
> -	vm_flags = calc_vm_prot_bits(prot);
> +	vm_prots = calc_vm_prot_bits(prot);
>  
>  	down_write(&current->mm->mmap_sem);
>  
> @@ -278,11 +278,11 @@ SYSCALL_DEFINE3(mprotect, unsigned long,
>  		prev = vma;
>  
>  	for (nstart = start ; ; ) {
> -		unsigned long newflags;
> +		unsigned long long newflags;
>  
>  		/* Here we know that  vma->vm_start <= nstart < vma->vm_end. */
>  
> -		newflags = vm_flags | (vma->vm_flags & ~(VM_READ | VM_WRITE | VM_EXEC));
> +		newflags = vm_prots | (vma->vm_flags & ~(VM_READ | VM_WRITE | VM_EXEC));
>  
>  		/* newflags >> 4 shift VM_MAY% in place of VM_% */
>  		if ((newflags & ~(newflags >> 4)) & (VM_READ | VM_WRITE | VM_EXEC)) {
> --- 2.6.32-rc1/mm/mremap.c	2009-09-28 00:28:41.000000000 +0100
> +++ ull_vm_flags/mm/mremap.c	2009-09-29 16:48:15.000000000 +0100
> @@ -169,7 +169,7 @@ static unsigned long move_vma(struct vm_
>  {
>  	struct mm_struct *mm = vma->vm_mm;
>  	struct vm_area_struct *new_vma;
> -	unsigned long vm_flags = vma->vm_flags;
> +	unsigned long long vm_flags = vma->vm_flags;
>  	unsigned long new_pgoff;
>  	unsigned long moved_len;
>  	unsigned long excess = 0;
> --- 2.6.32-rc1/mm/nommu.c	2009-09-28 00:28:41.000000000 +0100
> +++ ull_vm_flags/mm/nommu.c	2009-09-29 16:48:15.000000000 +0100
> @@ -134,7 +134,7 @@ int __get_user_pages(struct task_struct
>  		     struct page **pages, struct vm_area_struct **vmas)
>  {
>  	struct vm_area_struct *vma;
> -	unsigned long vm_flags;
> +	unsigned long long vm_flags;
>  	int i;
>  
>  	/* calculate required read or write permissions.
> @@ -987,12 +987,12 @@ static int validate_mmap_request(struct
>   * we've determined that we can make the mapping, now translate what we
>   * now know into VMA flags
>   */
> -static unsigned long determine_vm_flags(struct file *file,
> -					unsigned long prot,
> -					unsigned long flags,
> -					unsigned long capabilities)
> +static unsigned long long determine_vm_flags(struct file *file,
> +					     unsigned long prot,
> +					     unsigned long flags,
> +					     unsigned long capabilities)
>  {
> -	unsigned long vm_flags;
> +	unsigned long long vm_flags;
>  
>  	vm_flags = calc_vm_prot_bits(prot) | calc_vm_flag_bits(flags);
>  	vm_flags |= VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC;
> @@ -1177,7 +1177,8 @@ unsigned long do_mmap_pgoff(struct file
>  	struct vm_area_struct *vma;
>  	struct vm_region *region;
>  	struct rb_node *rb;
> -	unsigned long capabilities, vm_flags, result;
> +	unsigned long long vm_flags;
> +	unsigned long capabilities, result;
>  	int ret;
>  
>  	kenter(",%lx,%lx,%lx,%lx,%lx", addr, len, prot, flags, pgoff);
> --- 2.6.32-rc1/mm/rmap.c	2009-09-28 00:28:41.000000000 +0100
> +++ ull_vm_flags/mm/rmap.c	2009-09-29 16:48:15.000000000 +0100
> @@ -340,7 +340,7 @@ int page_mapped_in_vma(struct page *page
>  static int page_referenced_one(struct page *page,
>  			       struct vm_area_struct *vma,
>  			       unsigned int *mapcount,
> -			       unsigned long *vm_flags)
> +			       unsigned long long *vm_flags)
>  {
>  	struct mm_struct *mm = vma->vm_mm;
>  	unsigned long address;
> @@ -375,7 +375,7 @@ static int page_referenced_one(struct pa
>  		 * mapping is already gone, the unmap path will have
>  		 * set PG_referenced or activated the page.
>  		 */
> -		if (likely(!VM_SequentialReadHint(vma)))
> +		if (likely(!(vma->vm_flags & VM_SEQ_READ)))
>  			referenced++;
>  	}
>  
> @@ -396,7 +396,7 @@ out:
>  
>  static int page_referenced_anon(struct page *page,
>  				struct mem_cgroup *mem_cont,
> -				unsigned long *vm_flags)
> +				unsigned long long *vm_flags)
>  {
>  	unsigned int mapcount;
>  	struct anon_vma *anon_vma;
> @@ -441,7 +441,7 @@ static int page_referenced_anon(struct p
>   */
>  static int page_referenced_file(struct page *page,
>  				struct mem_cgroup *mem_cont,
> -				unsigned long *vm_flags)
> +				unsigned long long *vm_flags)
>  {
>  	unsigned int mapcount;
>  	struct address_space *mapping = page->mapping;
> @@ -504,7 +504,7 @@ static int page_referenced_file(struct p
>  int page_referenced(struct page *page,
>  		    int is_locked,
>  		    struct mem_cgroup *mem_cont,
> -		    unsigned long *vm_flags)
> +		    unsigned long long *vm_flags)
>  {
>  	int referenced = 0;
>  
> --- 2.6.32-rc1/mm/shmem.c	2009-09-28 00:28:41.000000000 +0100
> +++ ull_vm_flags/mm/shmem.c	2009-09-29 16:48:15.000000000 +0100
> @@ -2683,7 +2683,7 @@ int shmem_zero_setup(struct vm_area_stru
>  	struct file *file;
>  	loff_t size = vma->vm_end - vma->vm_start;
>  
> -	file = shmem_file_setup("dev/zero", size, vma->vm_flags);
> +	file = shmem_file_setup("dev/zero", size, (unsigned long)vma->vm_flags);
>  	if (IS_ERR(file))
>  		return PTR_ERR(file);
>  
> --- 2.6.32-rc1/mm/vmscan.c	2009-09-28 00:28:41.000000000 +0100
> +++ ull_vm_flags/mm/vmscan.c	2009-09-29 16:48:15.000000000 +0100
> @@ -581,7 +581,7 @@ static unsigned long shrink_page_list(st
>  	struct pagevec freed_pvec;
>  	int pgactivate = 0;
>  	unsigned long nr_reclaimed = 0;
> -	unsigned long vm_flags;
> +	unsigned long long vm_flags;
>  
>  	cond_resched();
>  
> @@ -1303,7 +1303,7 @@ static void shrink_active_list(unsigned
>  {
>  	unsigned long nr_taken;
>  	unsigned long pgscanned;
> -	unsigned long vm_flags;
> +	unsigned long long vm_flags;
>  	LIST_HEAD(l_hold);	/* The pages which were snipped off */
>  	LIST_HEAD(l_active);
>  	LIST_HEAD(l_inactive);
> 

Re: No more bits in vm_area_struct's vm_flags.

[KAMEZAWA Hiroyuki]

On Fri, 2 Oct 2009 09:42:38 +0900
KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> wrote:

> > > 
> > > (2) All vm macros should be defined with ULL suffix. for supporing ~ 
> > > ==
> > > vm_flags           30 arch/x86/mm/hugetlbpage.c 	unsigned long vm_flags = vma->vm_flags & ~VM_LOCKED;
> > > 
> > > (3) vma_merge()'s vm_flags should be ULL.
> > 
> > At first I thought you'd saved me a lot of embarrassment by mentioning
> > those ULL suffixes, I hadn't put them in.  But after a quick test of
> > what I thought was going to show a problem without them, no problem.
> > Please would you send me a test program which demonstrates the need
> > for all those ULLs?
> > 
> Ah, I'm sorry if I misunderstand C's rule. 
> 
> There are some places which use ~.
> like
> 	vm_flags = vma->vm_flags & ~(VM_LOCKED);
> 
> ~VM_LOCKED is 
> 	0xffffdfff or 0xffffffffffffdffff ?
> 
> Is my concern.
> 
> I tried following function on my old x86 box
> ==
> #define FLAG    (0x20)
> 
> int foo(unsigned long long x)
> {
>         return x & ~FLAG;
> }
> ==
> (returning "int" as "bool")
> 
> compile this with gcc -S -O2 (gcc's version is 4.0)
> ==
> foo:
>         pushl   %ebp
>         movl    %esp, %ebp
>         movl    8(%ebp), %eax
>         andl    $-33, %eax
>         leave
>         ret
> ==
> Them, it seems higher bits are ignored for returning bool.
> 
Sigh, I seems I don't undestand C language yet..

This one
==
#define FLAG    (0x20ULL)

int foo(unsigned long long x)
{
        return (x & ~FLAG);
}
==
is compiled as
==
foo:
        pushl   %ebp
        movl    %esp, %ebp
        movl    8(%ebp), %eax
        andl    $-33, %eax
        leave
        ret
==
ULL suffix makes no difference ;)

This one
==
#define FLAG    (0x20)

int foo(unsigned long long x)
{
        if (x & ~FLAG)
                return 1;
        return 0;
}

==
foo:
        pushl   %ebp
        movl    %esp, %ebp
        movl    8(%ebp), %eax
        movl    12(%ebp), %edx
        andl    $-33, %eax
        orl     %edx, %eax
        setne   %al
        movzbl  %al, %eax
        leave
        ret

seems good.


Hmm. sorry for noise. Maybe I don't understand C's cast rules.

-Kame

Re: No more bits in vm_area_struct's vm_flags.

[KAMEZAWA Hiroyuki]

On Fri, 2 Oct 2009 10:37:55 +0900
KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> wrote:
> This one
> ==
> #define FLAG    (0x20)
> 
> int foo(unsigned long long x)
> {
>         if (x & ~FLAG)
>                 return 1;
>         return 0;
> }
> 
Then, !!(x & ~FLAG) is necessary...i see (>_<
Maybe no problem..

Thanks,
-Kame