diff mbox series

[11/11] xfs: Remove kmem_alloc_{io, large} and kmem_zalloc_large

Message ID 20191113142335.1045631-12-cmaiolino@redhat.com (mailing list archive)
State Deferred, archived
Headers show
Series Use generic memory API instead of a custom one | expand

Commit Message

Carlos Maiolino Nov. 13, 2019, 2:23 p.m. UTC
Getting rid of these functions, is a bit more complicated, giving the
fact they use a vmalloc fallback, and (in case of _io version) uses an
alignment check, so they have their useness.

Instead of keeping both of them, I think sharing the same function for
both cases is a more interesting idea, giving the fact they both have
the same purpose, with the only difference being the alignment check,
which can be selected by using a flag.

Signed-off-by: Carlos Maiolino <cmaiolino@redhat.com>
---
 fs/xfs/kmem.c                 | 39 +++++++++++------------------------
 fs/xfs/kmem.h                 | 10 +--------
 fs/xfs/libxfs/xfs_attr_leaf.c |  2 +-
 fs/xfs/scrub/attr.c           |  2 +-
 fs/xfs/scrub/symlink.c        |  3 ++-
 fs/xfs/xfs_acl.c              |  3 ++-
 fs/xfs/xfs_buf.c              |  4 ++--
 fs/xfs/xfs_ioctl.c            |  8 ++++---
 fs/xfs/xfs_ioctl32.c          |  3 ++-
 fs/xfs/xfs_log.c              |  5 +++--
 fs/xfs/xfs_log_cil.c          |  2 +-
 fs/xfs/xfs_log_recover.c      |  4 ++--
 fs/xfs/xfs_rtalloc.c          |  3 ++-
 13 files changed, 36 insertions(+), 52 deletions(-)

Comments

Darrick J. Wong Nov. 13, 2019, 6:23 p.m. UTC | #1
On Wed, Nov 13, 2019 at 03:23:35PM +0100, Carlos Maiolino wrote:
> Getting rid of these functions, is a bit more complicated, giving the
> fact they use a vmalloc fallback, and (in case of _io version) uses an
> alignment check, so they have their useness.
> 
> Instead of keeping both of them, I think sharing the same function for
> both cases is a more interesting idea, giving the fact they both have
> the same purpose, with the only difference being the alignment check,
> which can be selected by using a flag.
> 
> Signed-off-by: Carlos Maiolino <cmaiolino@redhat.com>
> ---
>  fs/xfs/kmem.c                 | 39 +++++++++++------------------------
>  fs/xfs/kmem.h                 | 10 +--------
>  fs/xfs/libxfs/xfs_attr_leaf.c |  2 +-
>  fs/xfs/scrub/attr.c           |  2 +-
>  fs/xfs/scrub/symlink.c        |  3 ++-
>  fs/xfs/xfs_acl.c              |  3 ++-
>  fs/xfs/xfs_buf.c              |  4 ++--
>  fs/xfs/xfs_ioctl.c            |  8 ++++---
>  fs/xfs/xfs_ioctl32.c          |  3 ++-
>  fs/xfs/xfs_log.c              |  5 +++--
>  fs/xfs/xfs_log_cil.c          |  2 +-
>  fs/xfs/xfs_log_recover.c      |  4 ++--
>  fs/xfs/xfs_rtalloc.c          |  3 ++-
>  13 files changed, 36 insertions(+), 52 deletions(-)
> 
> diff --git a/fs/xfs/kmem.c b/fs/xfs/kmem.c
> index 44145293cfc9..bb4990970647 100644
> --- a/fs/xfs/kmem.c
> +++ b/fs/xfs/kmem.c
> @@ -8,40 +8,25 @@
>  #include "xfs_message.h"
>  #include "xfs_trace.h"
>  
> -/*
> - * Same as kmem_alloc_large, except we guarantee the buffer returned is aligned
> - * to the @align_mask. We only guarantee alignment up to page size, we'll clamp
> - * alignment at page size if it is larger. vmalloc always returns a PAGE_SIZE
> - * aligned region.
> - */
>  void *
> -kmem_alloc_io(size_t size, int align_mask, gfp_t flags)
> +xfs_kmem_alloc(size_t size, gfp_t flags, bool align, int align_mask)

A boolean for the align /and/ an alignment mask?  Yuck.

I think I'd rather have:

void *
kmem_alloc(
	size_t		size,
	gfp_t		flags,
	unsigned int	align_mask)
{
	... allocation logic ...
}

and in kmem.h:

static inline void *
kmem_alloc_io(
	size_t		size,
	gfp_t		flags,
	unsigned int	align_mask)
{
	trace_kmem_alloc_io(size, flags, align_mask, _RET_IP_);
	return kmem_alloc(size, flags, align_mask);
}

static inline void *
kmem_alloc_large(
	size_t		size,
	gfp_t		flags)
{
	trace_kmem_alloc_large(size, flags, _RET_IP_);
	return kmem_alloc(size, flags, 0);
}

because there's little sense in cluttering up the large allocation call
sites with "false, 0" parameters that make no sense for an incore memory
allocation that has no alignment requirements.

--D

>  {
>  	void	*ptr;
>  
> -	trace_kmem_alloc_io(size, flags, _RET_IP_);
> -
> -	if (WARN_ON_ONCE(align_mask >= PAGE_SIZE))
> -		align_mask = PAGE_SIZE - 1;
> -
>  	ptr = kmalloc(size, flags | __GFP_RETRY_MAYFAIL);
>  	if (ptr) {
> -		if (!((uintptr_t)ptr & align_mask))
> +		if (align) {
> +			trace_kmem_alloc_io(size, flags, _RET_IP_);
> +			if (WARN_ON_ONCE(align_mask >= PAGE_SIZE))
> +				align_mask = PAGE_SIZE - 1;
> +
> +			if (!((uintptr_t)ptr & align_mask))
> +				return ptr;
> +			kfree(ptr);
> +		} else {
> +			trace_kmem_alloc_large(size, flags, _RET_IP_);
>  			return ptr;
> -		kfree(ptr);
> +		}
>  	}
>  	return __vmalloc(size, flags | __GFP_NOFAIL, PAGE_KERNEL);
>  }
> -
> -void *
> -kmem_alloc_large(size_t size, gfp_t flags)
> -{
> -	void	*ptr;
> -
> -	trace_kmem_alloc_large(size, flags, _RET_IP_);
> -
> -	ptr = kmalloc(size, flags | __GFP_RETRY_MAYFAIL);
> -	if (ptr)
> -		return ptr;
> -	return __vmalloc(size, flags | __GFP_NOFAIL, PAGE_KERNEL);
> -}
> diff --git a/fs/xfs/kmem.h b/fs/xfs/kmem.h
> index 791e770be0eb..ee4c0152cdeb 100644
> --- a/fs/xfs/kmem.h
> +++ b/fs/xfs/kmem.h
> @@ -15,20 +15,12 @@
>   * General memory allocation interfaces
>   */
>  
> -extern void *kmem_alloc_io(size_t size, int align_mask, gfp_t flags);
> -extern void *kmem_alloc_large(size_t size, gfp_t);
> +extern void *xfs_kmem_alloc(size_t, gfp_t, bool, int);
>  static inline void  kmem_free(const void *ptr)
>  {
>  	kvfree(ptr);
>  }
>  
> -
> -static inline void *
> -kmem_zalloc_large(size_t size, gfp_t flags)
> -{
> -	return kmem_alloc_large(size, flags | __GFP_ZERO);
> -}
> -
>  /*
>   * Zone interfaces
>   */
> diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c
> index d3f872460ea6..eeb90f63cf2e 100644
> --- a/fs/xfs/libxfs/xfs_attr_leaf.c
> +++ b/fs/xfs/libxfs/xfs_attr_leaf.c
> @@ -479,7 +479,7 @@ xfs_attr_copy_value(
>  	}
>  
>  	if (args->op_flags & XFS_DA_OP_ALLOCVAL) {
> -		args->value = kmem_alloc_large(valuelen, GFP_KERNEL);
> +		args->value = xfs_kmem_alloc(valuelen, GFP_KERNEL, false, 0);
>  		if (!args->value)
>  			return -ENOMEM;
>  	}
> diff --git a/fs/xfs/scrub/attr.c b/fs/xfs/scrub/attr.c
> index bc09c46f4ff2..90239b902b47 100644
> --- a/fs/xfs/scrub/attr.c
> +++ b/fs/xfs/scrub/attr.c
> @@ -57,7 +57,7 @@ xchk_setup_xattr_buf(
>  	 * Don't zero the buffer upon allocation to avoid runtime overhead.
>  	 * All users must be careful never to read uninitialized contents.
>  	 */
> -	ab = kmem_alloc_large(sizeof(*ab) + sz, flags);
> +	ab = xfs_kmem_alloc(sizeof(*ab) + sz, flags, false, 0);
>  	if (!ab)
>  		return -ENOMEM;
>  
> diff --git a/fs/xfs/scrub/symlink.c b/fs/xfs/scrub/symlink.c
> index 5641ae512c9e..78f6d0dd8f2e 100644
> --- a/fs/xfs/scrub/symlink.c
> +++ b/fs/xfs/scrub/symlink.c
> @@ -22,7 +22,8 @@ xchk_setup_symlink(
>  	struct xfs_inode	*ip)
>  {
>  	/* Allocate the buffer without the inode lock held. */
> -	sc->buf = kmem_zalloc_large(XFS_SYMLINK_MAXLEN + 1, 0);
> +	sc->buf = xfs_kmem_alloc(XFS_SYMLINK_MAXLEN + 1,
> +				 GFP_KERNEL | __GFP_ZERO, false, 0);
>  	if (!sc->buf)
>  		return -ENOMEM;
>  
> diff --git a/fs/xfs/xfs_acl.c b/fs/xfs/xfs_acl.c
> index 91693fce34a8..988598e4e07c 100644
> --- a/fs/xfs/xfs_acl.c
> +++ b/fs/xfs/xfs_acl.c
> @@ -186,7 +186,8 @@ __xfs_set_acl(struct inode *inode, struct posix_acl *acl, int type)
>  		struct xfs_acl *xfs_acl;
>  		int len = XFS_ACL_MAX_SIZE(ip->i_mount);
>  
> -		xfs_acl = kmem_zalloc_large(len, 0);
> +		xfs_acl = xfs_kmem_alloc(len, GFP_KERNEL | __GFP_ZERO,
> +					 false, 0);
>  		if (!xfs_acl)
>  			return -ENOMEM;
>  
> diff --git a/fs/xfs/xfs_buf.c b/fs/xfs/xfs_buf.c
> index 678e024f7f1c..b36e4c4d3b9a 100644
> --- a/fs/xfs/xfs_buf.c
> +++ b/fs/xfs/xfs_buf.c
> @@ -361,8 +361,8 @@ xfs_buf_allocate_memory(
>  	size = BBTOB(bp->b_length);
>  	if (size < PAGE_SIZE) {
>  		int align_mask = xfs_buftarg_dma_alignment(bp->b_target);
> -		bp->b_addr = kmem_alloc_io(size, align_mask,
> -					   GFP_NOFS | __GFP_ZERO);
> +		bp->b_addr = xfs_kmem_alloc(size, GFP_NOFS | __GFP_ZERO, true,
> +					    align_mask);
>  		if (!bp->b_addr) {
>  			/* low memory - use alloc_page loop instead */
>  			goto use_alloc_page;
> diff --git a/fs/xfs/xfs_ioctl.c b/fs/xfs/xfs_ioctl.c
> index 364961c23cd0..72e26b7ac48f 100644
> --- a/fs/xfs/xfs_ioctl.c
> +++ b/fs/xfs/xfs_ioctl.c
> @@ -398,7 +398,8 @@ xfs_attrlist_by_handle(
>  	if (IS_ERR(dentry))
>  		return PTR_ERR(dentry);
>  
> -	kbuf = kmem_zalloc_large(al_hreq.buflen, 0);
> +	kbuf = xfs_kmem_alloc(al_hreq.buflen, GFP_KERNEL | __GFP_ZERO,
> +			      false, 0);
>  	if (!kbuf)
>  		goto out_dput;
>  
> @@ -436,7 +437,7 @@ xfs_attrmulti_attr_get(
>  
>  	if (*len > XFS_XATTR_SIZE_MAX)
>  		return -EINVAL;
> -	kbuf = kmem_zalloc_large(*len, 0);
> +	kbuf = xfs_kmem_alloc(*len, GFP_KERNEL | __GFP_ZERO, false, 0);
>  	if (!kbuf)
>  		return -ENOMEM;
>  
> @@ -1756,7 +1757,8 @@ xfs_ioc_getbmap(
>  	if (bmx.bmv_count > ULONG_MAX / recsize)
>  		return -ENOMEM;
>  
> -	buf = kmem_zalloc_large(bmx.bmv_count * sizeof(*buf), 0);
> +	buf = xfs_kmem_alloc(bmx.bmv_count * sizeof(*buf),
> +			     GFP_KERNEL | __GFP_ZERO, false, 0);
>  	if (!buf)
>  		return -ENOMEM;
>  
> diff --git a/fs/xfs/xfs_ioctl32.c b/fs/xfs/xfs_ioctl32.c
> index 3c0d518e1039..99886b1ba319 100644
> --- a/fs/xfs/xfs_ioctl32.c
> +++ b/fs/xfs/xfs_ioctl32.c
> @@ -381,7 +381,8 @@ xfs_compat_attrlist_by_handle(
>  		return PTR_ERR(dentry);
>  
>  	error = -ENOMEM;
> -	kbuf = kmem_zalloc_large(al_hreq.buflen, 0);
> +	kbuf = xfs_kmem_alloc(al_hreq.buflen, GFP_KERNEL | __GFP_ZERO,
> +			      false, 0);
>  	if (!kbuf)
>  		goto out_dput;
>  
> diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
> index dd65fdabf50e..c5e26080262c 100644
> --- a/fs/xfs/xfs_log.c
> +++ b/fs/xfs/xfs_log.c
> @@ -1491,8 +1491,9 @@ xlog_alloc_log(
>  		iclog->ic_prev = prev_iclog;
>  		prev_iclog = iclog;
>  
> -		iclog->ic_data = kmem_alloc_io(log->l_iclog_size, align_mask,
> -					       GFP_KERNEL | __GFP_ZERO);
> +		iclog->ic_data = xfs_kmem_alloc(log->l_iclog_size,
> +					       GFP_KERNEL | __GFP_ZERO,
> +					       true, align_mask);
>  		if (!iclog->ic_data)
>  			goto out_free_iclog;
>  #ifdef DEBUG
> diff --git a/fs/xfs/xfs_log_cil.c b/fs/xfs/xfs_log_cil.c
> index 9250b6b2f0fd..2585dbf653cc 100644
> --- a/fs/xfs/xfs_log_cil.c
> +++ b/fs/xfs/xfs_log_cil.c
> @@ -186,7 +186,7 @@ xlog_cil_alloc_shadow_bufs(
>  			 */
>  			kmem_free(lip->li_lv_shadow);
>  
> -			lv = kmem_alloc_large(buf_size, GFP_NOFS);
> +			lv = xfs_kmem_alloc(buf_size, GFP_NOFS, false, 0);
>  			memset(lv, 0, xlog_cil_iovec_space(niovecs));
>  
>  			lv->lv_item = lip;
> diff --git a/fs/xfs/xfs_log_recover.c b/fs/xfs/xfs_log_recover.c
> index 76b99ebdfcd9..3eb23f71a415 100644
> --- a/fs/xfs/xfs_log_recover.c
> +++ b/fs/xfs/xfs_log_recover.c
> @@ -127,8 +127,8 @@ xlog_alloc_buffer(
>  	if (nbblks > 1 && log->l_sectBBsize > 1)
>  		nbblks += log->l_sectBBsize;
>  	nbblks = round_up(nbblks, log->l_sectBBsize);
> -	return kmem_alloc_io(BBTOB(nbblks), align_mask,
> -			     GFP_KERNEL | __GFP_ZERO);
> +	return xfs_kmem_alloc(BBTOB(nbblks), GFP_KERNEL | __GFP_ZERO, true,
> +			      align_mask);
>  }
>  
>  /*
> diff --git a/fs/xfs/xfs_rtalloc.c b/fs/xfs/xfs_rtalloc.c
> index 1875484123d7..b2fa5f1a6acb 100644
> --- a/fs/xfs/xfs_rtalloc.c
> +++ b/fs/xfs/xfs_rtalloc.c
> @@ -864,7 +864,8 @@ xfs_alloc_rsum_cache(
>  	 * lower bound on the minimum level with any free extents. We can
>  	 * continue without the cache if it couldn't be allocated.
>  	 */
> -	mp->m_rsum_cache = kmem_zalloc_large(rbmblocks, 0);
> +	mp->m_rsum_cache = xfs_kmem_alloc(rbmblocks, GFP_KERNEL | __GFP_ZERO,
> +					  false, 0);
>  	if (!mp->m_rsum_cache)
>  		xfs_warn(mp, "could not allocate realtime summary cache");
>  }
> -- 
> 2.23.0
>
Dave Chinner Nov. 13, 2019, 8:06 p.m. UTC | #2
On Wed, Nov 13, 2019 at 10:23:43AM -0800, Darrick J. Wong wrote:
> On Wed, Nov 13, 2019 at 03:23:35PM +0100, Carlos Maiolino wrote:
> > Getting rid of these functions, is a bit more complicated, giving the
> > fact they use a vmalloc fallback, and (in case of _io version) uses an
> > alignment check, so they have their useness.
> > 
> > Instead of keeping both of them, I think sharing the same function for
> > both cases is a more interesting idea, giving the fact they both have
> > the same purpose, with the only difference being the alignment check,
> > which can be selected by using a flag.
> > 
> > Signed-off-by: Carlos Maiolino <cmaiolino@redhat.com>
> > ---
> >  fs/xfs/kmem.c                 | 39 +++++++++++------------------------
> >  fs/xfs/kmem.h                 | 10 +--------
> >  fs/xfs/libxfs/xfs_attr_leaf.c |  2 +-
> >  fs/xfs/scrub/attr.c           |  2 +-
> >  fs/xfs/scrub/symlink.c        |  3 ++-
> >  fs/xfs/xfs_acl.c              |  3 ++-
> >  fs/xfs/xfs_buf.c              |  4 ++--
> >  fs/xfs/xfs_ioctl.c            |  8 ++++---
> >  fs/xfs/xfs_ioctl32.c          |  3 ++-
> >  fs/xfs/xfs_log.c              |  5 +++--
> >  fs/xfs/xfs_log_cil.c          |  2 +-
> >  fs/xfs/xfs_log_recover.c      |  4 ++--
> >  fs/xfs/xfs_rtalloc.c          |  3 ++-
> >  13 files changed, 36 insertions(+), 52 deletions(-)
> > 
> > diff --git a/fs/xfs/kmem.c b/fs/xfs/kmem.c
> > index 44145293cfc9..bb4990970647 100644
> > --- a/fs/xfs/kmem.c
> > +++ b/fs/xfs/kmem.c
> > @@ -8,40 +8,25 @@
> >  #include "xfs_message.h"
> >  #include "xfs_trace.h"
> >  
> > -/*
> > - * Same as kmem_alloc_large, except we guarantee the buffer returned is aligned
> > - * to the @align_mask. We only guarantee alignment up to page size, we'll clamp
> > - * alignment at page size if it is larger. vmalloc always returns a PAGE_SIZE
> > - * aligned region.
> > - */
> >  void *
> > -kmem_alloc_io(size_t size, int align_mask, gfp_t flags)
> > +xfs_kmem_alloc(size_t size, gfp_t flags, bool align, int align_mask)
> 
> A boolean for the align /and/ an alignment mask?  Yuck.
> 
> I think I'd rather have:
> 
> void *
> kmem_alloc(
> 	size_t		size,
> 	gfp_t		flags,
> 	unsigned int	align_mask)
> {
> 	... allocation logic ...
> }

If you avoid changing the order of the flags/alignmask parameters,
most of the churn in this patch goes away.

> 
> and in kmem.h:
> 
> static inline void *
> kmem_alloc_io(
> 	size_t		size,
> 	gfp_t		flags,
> 	unsigned int	align_mask)
> {
> 	trace_kmem_alloc_io(size, flags, align_mask, _RET_IP_);
> 	return kmem_alloc(size, flags, align_mask);
> }

This should be able to go away soon, because the heap allocator will
guarantee alignment soon. That means kmem.c is a single function,
and kmem.h is a single function. I'd be looking to move the two
helper functions into some other utility file at that point
(fsops?)...

ANother question: how much work is there to be done on the userspace
side of things?

> >   */
> >  
> > -extern void *kmem_alloc_io(size_t size, int align_mask, gfp_t flags);
> > -extern void *kmem_alloc_large(size_t size, gfp_t);
> > +extern void *xfs_kmem_alloc(size_t, gfp_t, bool, int);
> >  static inline void  kmem_free(const void *ptr)
> >  {
> >  	kvfree(ptr);
> >  }

Didn't an earlier patch get rid of kmem_free(), or am I just
imagining this? Seems silly to leave this behind, now that the
only place that needs kvfree() is the callers to kmem_alloc_io and
kmem_alloc_large...

Cheers,

Dave.
Eric Sandeen Nov. 13, 2019, 8:43 p.m. UTC | #3
On 11/13/19 2:06 PM, Dave Chinner wrote:
>> and in kmem.h:
>>
>> static inline void *
>> kmem_alloc_io(
>> 	size_t		size,
>> 	gfp_t		flags,
>> 	unsigned int	align_mask)
>> {
>> 	trace_kmem_alloc_io(size, flags, align_mask, _RET_IP_);
>> 	return kmem_alloc(size, flags, align_mask);
>> }
> This should be able to go away soon, because the heap allocator will
> guarantee alignment soon. That means kmem.c is a single function,
> and kmem.h is a single function. I'd be looking to move the two
> helper functions into some other utility file at that point
> (fsops?)...
> 
> ANother question: how much work is there to be done on the userspace
> side of things?

Right now userspace only uses kmem_alloc, with the 5.4 sync there is one
call to kmem_alloc_large which I just #defined to kmem_alloc in my patch
stack.

kmem_alloc_large is in libxfs so it'd just get fixed up if/when this patch is
merged across.

-Eric
Carlos Maiolino Nov. 14, 2019, 9:46 a.m. UTC | #4
> ANother question: how much work is there to be done on the userspace
> side of things?
> 
> > >   */
> > >  
> > > -extern void *kmem_alloc_io(size_t size, int align_mask, gfp_t flags);
> > > -extern void *kmem_alloc_large(size_t size, gfp_t);
> > > +extern void *xfs_kmem_alloc(size_t, gfp_t, bool, int);
> > >  static inline void  kmem_free(const void *ptr)
> > >  {
> > >  	kvfree(ptr);
> > >  }
> 
> Didn't an earlier patch get rid of kmem_free(), or am I just
> imagining this? Seems silly to leave this behind, now that the
> only place that needs kvfree() is the callers to kmem_alloc_io and
> kmem_alloc_large...

Yes, it's on my roadmap, I just wanted to work on the other wrappers first and
send it later when I had kmem_alloc_{io, large}() worked out. I'll make sure to
convert this one on the V2.

> 
> Cheers,
> 
> Dave.
> -- 
> Dave Chinner
> david@fromorbit.com
>
diff mbox series

Patch

diff --git a/fs/xfs/kmem.c b/fs/xfs/kmem.c
index 44145293cfc9..bb4990970647 100644
--- a/fs/xfs/kmem.c
+++ b/fs/xfs/kmem.c
@@ -8,40 +8,25 @@ 
 #include "xfs_message.h"
 #include "xfs_trace.h"
 
-/*
- * Same as kmem_alloc_large, except we guarantee the buffer returned is aligned
- * to the @align_mask. We only guarantee alignment up to page size, we'll clamp
- * alignment at page size if it is larger. vmalloc always returns a PAGE_SIZE
- * aligned region.
- */
 void *
-kmem_alloc_io(size_t size, int align_mask, gfp_t flags)
+xfs_kmem_alloc(size_t size, gfp_t flags, bool align, int align_mask)
 {
 	void	*ptr;
 
-	trace_kmem_alloc_io(size, flags, _RET_IP_);
-
-	if (WARN_ON_ONCE(align_mask >= PAGE_SIZE))
-		align_mask = PAGE_SIZE - 1;
-
 	ptr = kmalloc(size, flags | __GFP_RETRY_MAYFAIL);
 	if (ptr) {
-		if (!((uintptr_t)ptr & align_mask))
+		if (align) {
+			trace_kmem_alloc_io(size, flags, _RET_IP_);
+			if (WARN_ON_ONCE(align_mask >= PAGE_SIZE))
+				align_mask = PAGE_SIZE - 1;
+
+			if (!((uintptr_t)ptr & align_mask))
+				return ptr;
+			kfree(ptr);
+		} else {
+			trace_kmem_alloc_large(size, flags, _RET_IP_);
 			return ptr;
-		kfree(ptr);
+		}
 	}
 	return __vmalloc(size, flags | __GFP_NOFAIL, PAGE_KERNEL);
 }
-
-void *
-kmem_alloc_large(size_t size, gfp_t flags)
-{
-	void	*ptr;
-
-	trace_kmem_alloc_large(size, flags, _RET_IP_);
-
-	ptr = kmalloc(size, flags | __GFP_RETRY_MAYFAIL);
-	if (ptr)
-		return ptr;
-	return __vmalloc(size, flags | __GFP_NOFAIL, PAGE_KERNEL);
-}
diff --git a/fs/xfs/kmem.h b/fs/xfs/kmem.h
index 791e770be0eb..ee4c0152cdeb 100644
--- a/fs/xfs/kmem.h
+++ b/fs/xfs/kmem.h
@@ -15,20 +15,12 @@ 
  * General memory allocation interfaces
  */
 
-extern void *kmem_alloc_io(size_t size, int align_mask, gfp_t flags);
-extern void *kmem_alloc_large(size_t size, gfp_t);
+extern void *xfs_kmem_alloc(size_t, gfp_t, bool, int);
 static inline void  kmem_free(const void *ptr)
 {
 	kvfree(ptr);
 }
 
-
-static inline void *
-kmem_zalloc_large(size_t size, gfp_t flags)
-{
-	return kmem_alloc_large(size, flags | __GFP_ZERO);
-}
-
 /*
  * Zone interfaces
  */
diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c
index d3f872460ea6..eeb90f63cf2e 100644
--- a/fs/xfs/libxfs/xfs_attr_leaf.c
+++ b/fs/xfs/libxfs/xfs_attr_leaf.c
@@ -479,7 +479,7 @@  xfs_attr_copy_value(
 	}
 
 	if (args->op_flags & XFS_DA_OP_ALLOCVAL) {
-		args->value = kmem_alloc_large(valuelen, GFP_KERNEL);
+		args->value = xfs_kmem_alloc(valuelen, GFP_KERNEL, false, 0);
 		if (!args->value)
 			return -ENOMEM;
 	}
diff --git a/fs/xfs/scrub/attr.c b/fs/xfs/scrub/attr.c
index bc09c46f4ff2..90239b902b47 100644
--- a/fs/xfs/scrub/attr.c
+++ b/fs/xfs/scrub/attr.c
@@ -57,7 +57,7 @@  xchk_setup_xattr_buf(
 	 * Don't zero the buffer upon allocation to avoid runtime overhead.
 	 * All users must be careful never to read uninitialized contents.
 	 */
-	ab = kmem_alloc_large(sizeof(*ab) + sz, flags);
+	ab = xfs_kmem_alloc(sizeof(*ab) + sz, flags, false, 0);
 	if (!ab)
 		return -ENOMEM;
 
diff --git a/fs/xfs/scrub/symlink.c b/fs/xfs/scrub/symlink.c
index 5641ae512c9e..78f6d0dd8f2e 100644
--- a/fs/xfs/scrub/symlink.c
+++ b/fs/xfs/scrub/symlink.c
@@ -22,7 +22,8 @@  xchk_setup_symlink(
 	struct xfs_inode	*ip)
 {
 	/* Allocate the buffer without the inode lock held. */
-	sc->buf = kmem_zalloc_large(XFS_SYMLINK_MAXLEN + 1, 0);
+	sc->buf = xfs_kmem_alloc(XFS_SYMLINK_MAXLEN + 1,
+				 GFP_KERNEL | __GFP_ZERO, false, 0);
 	if (!sc->buf)
 		return -ENOMEM;
 
diff --git a/fs/xfs/xfs_acl.c b/fs/xfs/xfs_acl.c
index 91693fce34a8..988598e4e07c 100644
--- a/fs/xfs/xfs_acl.c
+++ b/fs/xfs/xfs_acl.c
@@ -186,7 +186,8 @@  __xfs_set_acl(struct inode *inode, struct posix_acl *acl, int type)
 		struct xfs_acl *xfs_acl;
 		int len = XFS_ACL_MAX_SIZE(ip->i_mount);
 
-		xfs_acl = kmem_zalloc_large(len, 0);
+		xfs_acl = xfs_kmem_alloc(len, GFP_KERNEL | __GFP_ZERO,
+					 false, 0);
 		if (!xfs_acl)
 			return -ENOMEM;
 
diff --git a/fs/xfs/xfs_buf.c b/fs/xfs/xfs_buf.c
index 678e024f7f1c..b36e4c4d3b9a 100644
--- a/fs/xfs/xfs_buf.c
+++ b/fs/xfs/xfs_buf.c
@@ -361,8 +361,8 @@  xfs_buf_allocate_memory(
 	size = BBTOB(bp->b_length);
 	if (size < PAGE_SIZE) {
 		int align_mask = xfs_buftarg_dma_alignment(bp->b_target);
-		bp->b_addr = kmem_alloc_io(size, align_mask,
-					   GFP_NOFS | __GFP_ZERO);
+		bp->b_addr = xfs_kmem_alloc(size, GFP_NOFS | __GFP_ZERO, true,
+					    align_mask);
 		if (!bp->b_addr) {
 			/* low memory - use alloc_page loop instead */
 			goto use_alloc_page;
diff --git a/fs/xfs/xfs_ioctl.c b/fs/xfs/xfs_ioctl.c
index 364961c23cd0..72e26b7ac48f 100644
--- a/fs/xfs/xfs_ioctl.c
+++ b/fs/xfs/xfs_ioctl.c
@@ -398,7 +398,8 @@  xfs_attrlist_by_handle(
 	if (IS_ERR(dentry))
 		return PTR_ERR(dentry);
 
-	kbuf = kmem_zalloc_large(al_hreq.buflen, 0);
+	kbuf = xfs_kmem_alloc(al_hreq.buflen, GFP_KERNEL | __GFP_ZERO,
+			      false, 0);
 	if (!kbuf)
 		goto out_dput;
 
@@ -436,7 +437,7 @@  xfs_attrmulti_attr_get(
 
 	if (*len > XFS_XATTR_SIZE_MAX)
 		return -EINVAL;
-	kbuf = kmem_zalloc_large(*len, 0);
+	kbuf = xfs_kmem_alloc(*len, GFP_KERNEL | __GFP_ZERO, false, 0);
 	if (!kbuf)
 		return -ENOMEM;
 
@@ -1756,7 +1757,8 @@  xfs_ioc_getbmap(
 	if (bmx.bmv_count > ULONG_MAX / recsize)
 		return -ENOMEM;
 
-	buf = kmem_zalloc_large(bmx.bmv_count * sizeof(*buf), 0);
+	buf = xfs_kmem_alloc(bmx.bmv_count * sizeof(*buf),
+			     GFP_KERNEL | __GFP_ZERO, false, 0);
 	if (!buf)
 		return -ENOMEM;
 
diff --git a/fs/xfs/xfs_ioctl32.c b/fs/xfs/xfs_ioctl32.c
index 3c0d518e1039..99886b1ba319 100644
--- a/fs/xfs/xfs_ioctl32.c
+++ b/fs/xfs/xfs_ioctl32.c
@@ -381,7 +381,8 @@  xfs_compat_attrlist_by_handle(
 		return PTR_ERR(dentry);
 
 	error = -ENOMEM;
-	kbuf = kmem_zalloc_large(al_hreq.buflen, 0);
+	kbuf = xfs_kmem_alloc(al_hreq.buflen, GFP_KERNEL | __GFP_ZERO,
+			      false, 0);
 	if (!kbuf)
 		goto out_dput;
 
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index dd65fdabf50e..c5e26080262c 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -1491,8 +1491,9 @@  xlog_alloc_log(
 		iclog->ic_prev = prev_iclog;
 		prev_iclog = iclog;
 
-		iclog->ic_data = kmem_alloc_io(log->l_iclog_size, align_mask,
-					       GFP_KERNEL | __GFP_ZERO);
+		iclog->ic_data = xfs_kmem_alloc(log->l_iclog_size,
+					       GFP_KERNEL | __GFP_ZERO,
+					       true, align_mask);
 		if (!iclog->ic_data)
 			goto out_free_iclog;
 #ifdef DEBUG
diff --git a/fs/xfs/xfs_log_cil.c b/fs/xfs/xfs_log_cil.c
index 9250b6b2f0fd..2585dbf653cc 100644
--- a/fs/xfs/xfs_log_cil.c
+++ b/fs/xfs/xfs_log_cil.c
@@ -186,7 +186,7 @@  xlog_cil_alloc_shadow_bufs(
 			 */
 			kmem_free(lip->li_lv_shadow);
 
-			lv = kmem_alloc_large(buf_size, GFP_NOFS);
+			lv = xfs_kmem_alloc(buf_size, GFP_NOFS, false, 0);
 			memset(lv, 0, xlog_cil_iovec_space(niovecs));
 
 			lv->lv_item = lip;
diff --git a/fs/xfs/xfs_log_recover.c b/fs/xfs/xfs_log_recover.c
index 76b99ebdfcd9..3eb23f71a415 100644
--- a/fs/xfs/xfs_log_recover.c
+++ b/fs/xfs/xfs_log_recover.c
@@ -127,8 +127,8 @@  xlog_alloc_buffer(
 	if (nbblks > 1 && log->l_sectBBsize > 1)
 		nbblks += log->l_sectBBsize;
 	nbblks = round_up(nbblks, log->l_sectBBsize);
-	return kmem_alloc_io(BBTOB(nbblks), align_mask,
-			     GFP_KERNEL | __GFP_ZERO);
+	return xfs_kmem_alloc(BBTOB(nbblks), GFP_KERNEL | __GFP_ZERO, true,
+			      align_mask);
 }
 
 /*
diff --git a/fs/xfs/xfs_rtalloc.c b/fs/xfs/xfs_rtalloc.c
index 1875484123d7..b2fa5f1a6acb 100644
--- a/fs/xfs/xfs_rtalloc.c
+++ b/fs/xfs/xfs_rtalloc.c
@@ -864,7 +864,8 @@  xfs_alloc_rsum_cache(
 	 * lower bound on the minimum level with any free extents. We can
 	 * continue without the cache if it couldn't be allocated.
 	 */
-	mp->m_rsum_cache = kmem_zalloc_large(rbmblocks, 0);
+	mp->m_rsum_cache = xfs_kmem_alloc(rbmblocks, GFP_KERNEL | __GFP_ZERO,
+					  false, 0);
 	if (!mp->m_rsum_cache)
 		xfs_warn(mp, "could not allocate realtime summary cache");
 }