MM: introduce memalloc_retry_wait()

Various places in the kernel - largely in filesystems - respond to a
memory allocation failure by looping around and re-trying.
Some of these cannot conveniently use __GFP_NOFAIL, for reasons such as:
 - a GFP_ATOMIC allocation, which __GFP_NOFAIL doesn't work on
 - a need to check for the process being signalled between failures
 - the possibility that other recovery actions could be performed
 - the allocation is quite deep in support code, and passing down an
   extra flag to say if __GFP_NOFAIL is wanted would be clumsy.

Many of these currently use congestion_wait() which (in almost all
cases) simply waits the given timeout - congestion isn't tracked for
most devices.

It isn't clear what the best delay is for loops, but it is clear that
the various filesystems shouldn't be responsible for choosing a timeout.

This patch introduces memalloc_retry_wait() with takes on that
responsibility.  Code that wants to retry a memory allocation can call
this function passing the GFP flags that were used.  It will wait
however is appropriate.

For now, it only considers the __GFP_DIRECT_RECLAIM and __GFP_NORETRY
flags.  If DIRECT_RECLAIM is set without GFP_NORETRY, then alloc_page
either made some reclaim progress, or waited for a while, before
failing.  So there is no need for much further waiting.
memalloc_retry_wait() will wait until the current jiffie ends.  If this
condition is not met, then alloc_page() won't have waited.  In that case
memalloc_retry_wait() waits about 200ms.  This is the delay that most
current loops uses.

linux/sched/mm.h needs to be included in some files now,
but linux/backing-dev.h does not.

Signed-off-by: NeilBrown <neilb@suse.de>
---

I could split this up by filesystems if people prefer that, but they we
would have to wait for the -mm patch to land before they could be
applied.  What do people think?  Can Andrew just collected acked-bys?

Thanks,
NeilBrown


 fs/ext4/extents.c        |  8 +++-----
 fs/ext4/inline.c         |  5 ++---
 fs/ext4/page-io.c        |  9 +++++----
 fs/f2fs/data.c           |  3 +--
 fs/f2fs/gc.c             |  5 ++---
 fs/f2fs/inode.c          |  4 ++--
 fs/f2fs/node.c           |  3 +--
 fs/f2fs/recovery.c       |  6 +++---
 fs/f2fs/segment.c        |  8 ++------
 fs/f2fs/super.c          |  4 +---
 fs/xfs/kmem.c            |  3 +--
 fs/xfs/xfs_buf.c         |  2 +-
 include/linux/sched/mm.h | 21 +++++++++++++++++++++
 net/sunrpc/svc_xprt.c    |  2 +-
 14 files changed, 46 insertions(+), 37 deletions(-)

On Wed, Nov 17, 2021 at 03:28:10PM +1100, NeilBrown wrote:
> 
> Various places in the kernel - largely in filesystems - respond to a
> memory allocation failure by looping around and re-trying.
.....
> diff --git a/include/linux/sched/mm.h b/include/linux/sched/mm.h
> index aca874d33fe6..f2f2a5b28808 100644
> --- a/include/linux/sched/mm.h
> +++ b/include/linux/sched/mm.h
> @@ -214,6 +214,27 @@ static inline void fs_reclaim_acquire(gfp_t gfp_mask) { }
>  static inline void fs_reclaim_release(gfp_t gfp_mask) { }
>  #endif
>  
> +/* Any memory-allocation retry loop should use
> + * memalloc_retry_wait(), and pass the flags for the most
> + * constrained allocation attempt that might have failed.
> + * This provides useful documentation of where loops are,
> + * and a central place to fine tune the waiting as the MM
> + * implementation changes.
> + */
> +static inline void memalloc_retry_wait(gfp_t gfp_flags)
> +{
> +	gfp_flags = current_gfp_context(gfp_flags);
> +	if ((gfp_flags & __GFP_DIRECT_RECLAIM) &&
> +	    !(gfp_flags & __GFP_NORETRY))
> +		/* Probably waited already, no need for much more */
> +		schedule_timeout_uninterruptible(1);
> +	else
> +		/* Probably didn't wait, and has now released a lock,
> +		 * so now is a good time to wait
> +		 */
> +		schedule_timeout_uninterruptible(HZ/50);
> +}

The existing congestion_wait() calls io_schedule_timeout() under
TASK_UNINTERRUPTIBLE conditions.

Does changing all these calls just to a plain
schedule_timeout_uninterruptible() make any difference to behaviour?
At least process accounting will appear different (uninterruptible
sleep instead of IO wait), and I suspect that the block plug
flushing in io_schedule() might be a good idea to retain for all the
filesystems that call this function from IO-related routines.

Cheers,

Dave.

Hi NeilBrown,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on jaegeuk-f2fs/dev-test]
[also build test ERROR on v5.16-rc1 next-20211118]
[cannot apply to tytso-ext4/dev xfs-linux/for-next]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/NeilBrown/MM-introduce-memalloc_retry_wait/20211117-122932
base:   https://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs.git dev-test
config: riscv-randconfig-r004-20211118 (attached as .config)
reproduce (this is a W=1 build):
        wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # install riscv cross compiling tool for clang build
        # apt-get install binutils-riscv64-linux-gnu
        # https://github.com/0day-ci/linux/commit/1da355dc212c5f6ade3a21d4a5b1cfaf6b48ff0f
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review NeilBrown/MM-introduce-memalloc_retry_wait/20211117-122932
        git checkout 1da355dc212c5f6ade3a21d4a5b1cfaf6b48ff0f
        # save the attached .config to linux build tree
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross W=1 ARCH=riscv 

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>

All errors (new ones prefixed by >>):

>> fs/f2fs/super.c:2417:5: error: implicit declaration of function 'memalloc_retry_wait' [-Werror,-Wimplicit-function-declaration]
                                   memalloc_retry_wait(GFP_NOFS);
                                   ^
   1 error generated.


vim +/memalloc_retry_wait +2417 fs/f2fs/super.c

  2389	
  2390	#ifdef CONFIG_QUOTA
  2391	/* Read data from quotafile */
  2392	static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
  2393				       size_t len, loff_t off)
  2394	{
  2395		struct inode *inode = sb_dqopt(sb)->files[type];
  2396		struct address_space *mapping = inode->i_mapping;
  2397		block_t blkidx = F2FS_BYTES_TO_BLK(off);
  2398		int offset = off & (sb->s_blocksize - 1);
  2399		int tocopy;
  2400		size_t toread;
  2401		loff_t i_size = i_size_read(inode);
  2402		struct page *page;
  2403		char *kaddr;
  2404	
  2405		if (off > i_size)
  2406			return 0;
  2407	
  2408		if (off + len > i_size)
  2409			len = i_size - off;
  2410		toread = len;
  2411		while (toread > 0) {
  2412			tocopy = min_t(unsigned long, sb->s_blocksize - offset, toread);
  2413	repeat:
  2414			page = read_cache_page_gfp(mapping, blkidx, GFP_NOFS);
  2415			if (IS_ERR(page)) {
  2416				if (PTR_ERR(page) == -ENOMEM) {
> 2417					memalloc_retry_wait(GFP_NOFS);
  2418					goto repeat;
  2419				}
  2420				set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
  2421				return PTR_ERR(page);
  2422			}
  2423	
  2424			lock_page(page);
  2425	
  2426			if (unlikely(page->mapping != mapping)) {
  2427				f2fs_put_page(page, 1);
  2428				goto repeat;
  2429			}
  2430			if (unlikely(!PageUptodate(page))) {
  2431				f2fs_put_page(page, 1);
  2432				set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
  2433				return -EIO;
  2434			}
  2435	
  2436			kaddr = kmap_atomic(page);
  2437			memcpy(data, kaddr + offset, tocopy);
  2438			kunmap_atomic(kaddr);
  2439			f2fs_put_page(page, 1);
  2440	
  2441			offset = 0;
  2442			toread -= tocopy;
  2443			data += tocopy;
  2444			blkidx++;
  2445		}
  2446		return len;
  2447	}
  2448	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org

Hi NeilBrown,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on jaegeuk-f2fs/dev-test]
[also build test ERROR on v5.16-rc1 next-20211118]
[cannot apply to xfs-linux/for-next]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/NeilBrown/MM-introduce-memalloc_retry_wait/20211117-122932
base:   https://git.kernel.org/pub/scm/linux/kernel/git/jaegeuk/f2fs.git dev-test
config: arc-randconfig-r043-20211117 (attached as .config)
compiler: arceb-elf-gcc (GCC) 11.2.0
reproduce (this is a W=1 build):
        wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # https://github.com/0day-ci/linux/commit/1da355dc212c5f6ade3a21d4a5b1cfaf6b48ff0f
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review NeilBrown/MM-introduce-memalloc_retry_wait/20211117-122932
        git checkout 1da355dc212c5f6ade3a21d4a5b1cfaf6b48ff0f
        # save the attached .config to linux build tree
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-11.2.0 make.cross ARCH=arc 

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>

All errors (new ones prefixed by >>):

   fs/f2fs/super.c: In function 'f2fs_quota_read':
>> fs/f2fs/super.c:2417:33: error: implicit declaration of function 'memalloc_retry_wait' [-Werror=implicit-function-declaration]
    2417 |                                 memalloc_retry_wait(GFP_NOFS);
         |                                 ^~~~~~~~~~~~~~~~~~~
   cc1: some warnings being treated as errors


vim +/memalloc_retry_wait +2417 fs/f2fs/super.c

  2389	
  2390	#ifdef CONFIG_QUOTA
  2391	/* Read data from quotafile */
  2392	static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
  2393				       size_t len, loff_t off)
  2394	{
  2395		struct inode *inode = sb_dqopt(sb)->files[type];
  2396		struct address_space *mapping = inode->i_mapping;
  2397		block_t blkidx = F2FS_BYTES_TO_BLK(off);
  2398		int offset = off & (sb->s_blocksize - 1);
  2399		int tocopy;
  2400		size_t toread;
  2401		loff_t i_size = i_size_read(inode);
  2402		struct page *page;
  2403		char *kaddr;
  2404	
  2405		if (off > i_size)
  2406			return 0;
  2407	
  2408		if (off + len > i_size)
  2409			len = i_size - off;
  2410		toread = len;
  2411		while (toread > 0) {
  2412			tocopy = min_t(unsigned long, sb->s_blocksize - offset, toread);
  2413	repeat:
  2414			page = read_cache_page_gfp(mapping, blkidx, GFP_NOFS);
  2415			if (IS_ERR(page)) {
  2416				if (PTR_ERR(page) == -ENOMEM) {
> 2417					memalloc_retry_wait(GFP_NOFS);
  2418					goto repeat;
  2419				}
  2420				set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
  2421				return PTR_ERR(page);
  2422			}
  2423	
  2424			lock_page(page);
  2425	
  2426			if (unlikely(page->mapping != mapping)) {
  2427				f2fs_put_page(page, 1);
  2428				goto repeat;
  2429			}
  2430			if (unlikely(!PageUptodate(page))) {
  2431				f2fs_put_page(page, 1);
  2432				set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
  2433				return -EIO;
  2434			}
  2435	
  2436			kaddr = kmap_atomic(page);
  2437			memcpy(data, kaddr + offset, tocopy);
  2438			kunmap_atomic(kaddr);
  2439			f2fs_put_page(page, 1);
  2440	
  2441			offset = 0;
  2442			toread -= tocopy;
  2443			data += tocopy;
  2444			blkidx++;
  2445		}
  2446		return len;
  2447	}
  2448	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org