Revision 289454ad26a2d752e04b07234a175feda9ec0f4e authored by Naohiro Aota on 05 January 2015, 16:01:03 UTC, committed by Chris Mason on 03 February 2015, 03:24:51 UTC
After submit_one_bio(), `bio' can go away. However submit_extent_page()
leave `bio' referable if submit_one_bio() failed (e.g. -ENOMEM on OOM).
It will cause invalid paging request when submit_extent_page() is called
next time.

I reproduced ENOMEM case with the following script (need
CONFIG_FAIL_PAGE_ALLOC, and CONFIG_FAULT_INJECTION_DEBUG_FS).

  #!/bin/bash

  dmesgout=dmesg.txt
  start=100000
  end=300000
  step=1000

  # btrfs options
  device=/dev/vdb1
  directory=/mnt/btrfs

  # fault-injection options
  percent=100
  times=3

  mkdir -p $directory || exit 1
  mount -o compress $device $directory || exit 1

  rm -f $directory/file || exit 1
  dd if=/dev/zero of=$directory/file bs=1M count=512 || exit 1

  for interval in `seq $start $step $end`; do
          dmesg -C
          echo 1 > /proc/sys/vm/drop_caches
          sync
          export FAILCMD_TYPE=fail_page_alloc
          ./failcmd.sh -p $percent -t $times -i $interval \
                  --ignore-gfp-highmem=N --ignore-gfp-wait=N --min-order=0 \
                  -- \
                  cat $directory/file > /dev/null
          dmesg > ${dmesgout}
          if grep -q BUG: ${dmesgout}; then
                  cat ${dmesgout}
                  exit 1
          fi
  done

  umount $directory
  exit 0

Signed-off-by: Naohiro Aota <naota@elisp.net>
Tested-by: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: Chris Mason <clm@fb.com>
1 parent de554a4
Raw File
svcshare.c
/*
 * linux/fs/lockd/svcshare.c
 *
 * Management of DOS shares.
 *
 * Copyright (C) 1996 Olaf Kirch <okir@monad.swb.de>
 */

#include <linux/time.h>
#include <linux/unistd.h>
#include <linux/string.h>
#include <linux/slab.h>

#include <linux/sunrpc/clnt.h>
#include <linux/sunrpc/svc.h>
#include <linux/lockd/lockd.h>
#include <linux/lockd/share.h>

static inline int
nlm_cmp_owner(struct nlm_share *share, struct xdr_netobj *oh)
{
	return share->s_owner.len == oh->len
	    && !memcmp(share->s_owner.data, oh->data, oh->len);
}

__be32
nlmsvc_share_file(struct nlm_host *host, struct nlm_file *file,
			struct nlm_args *argp)
{
	struct nlm_share	*share;
	struct xdr_netobj	*oh = &argp->lock.oh;
	u8			*ohdata;

	for (share = file->f_shares; share; share = share->s_next) {
		if (share->s_host == host && nlm_cmp_owner(share, oh))
			goto update;
		if ((argp->fsm_access & share->s_mode)
		 || (argp->fsm_mode   & share->s_access ))
			return nlm_lck_denied;
	}

	share = kmalloc(sizeof(*share) + oh->len,
						GFP_KERNEL);
	if (share == NULL)
		return nlm_lck_denied_nolocks;

	/* Copy owner handle */
	ohdata = (u8 *) (share + 1);
	memcpy(ohdata, oh->data, oh->len);

	share->s_file	    = file;
	share->s_host       = host;
	share->s_owner.data = ohdata;
	share->s_owner.len  = oh->len;
	share->s_next       = file->f_shares;
	file->f_shares      = share;

update:
	share->s_access = argp->fsm_access;
	share->s_mode   = argp->fsm_mode;
	return nlm_granted;
}

/*
 * Delete a share.
 */
__be32
nlmsvc_unshare_file(struct nlm_host *host, struct nlm_file *file,
			struct nlm_args *argp)
{
	struct nlm_share	*share, **shpp;
	struct xdr_netobj	*oh = &argp->lock.oh;

	for (shpp = &file->f_shares; (share = *shpp) != NULL;
					shpp = &share->s_next) {
		if (share->s_host == host && nlm_cmp_owner(share, oh)) {
			*shpp = share->s_next;
			kfree(share);
			return nlm_granted;
		}
	}

	/* X/Open spec says return success even if there was no
	 * corresponding share. */
	return nlm_granted;
}

/*
 * Traverse all shares for a given file, and delete
 * those owned by the given (type of) host
 */
void nlmsvc_traverse_shares(struct nlm_host *host, struct nlm_file *file,
		nlm_host_match_fn_t match)
{
	struct nlm_share	*share, **shpp;

	shpp = &file->f_shares;
	while ((share = *shpp) !=  NULL) {
		if (match(share->s_host, host)) {
			*shpp = share->s_next;
			kfree(share);
			continue;
		}
		shpp = &share->s_next;
	}
}
back to top