2017-03-27 5:36 GMT+02:00 NeilBrown <neilb@xxxxxxxx>:
> In rare circumstances, the short period that *hot_remove_disk()
> waits isn't long enough to IO to complete. This particularly happens
> when a device is failing and many retries are still happening.
>
> We don't want to increase the normal wait time for "mdadm --remove"
> as that might be use just to test if a device is active or not, and a
> delay would be problematic.
> So allow "--force" to mean that mdadm should try extra hard for a
> --remove to complete, waiting up to 5 seconds.
>
> Note that this patch fixes a comment which claim the previous
> wait time was half a second, where it was really 50msec.
>
> Signed-off-by: NeilBrown <neilb@xxxxxxxx>
> ---
> Grow.c | 2 +-
> Manage.c | 10 +++++-----
> mdadm.h | 4 ++--
> util.c | 10 +++++-----
> 4 files changed, 13 insertions(+), 13 deletions(-)
>
> diff --git a/Grow.c b/Grow.c
> index 218a70649c1f..e22661ca3dff 100755
> --- a/Grow.c
> +++ b/Grow.c
> @@ -2749,7 +2749,7 @@ static int impose_level(int fd, int level, char *devname, int verbose)
> continue;
> ioctl(fd, SET_DISK_FAULTY,
> makedev(disk.major, disk.minor));
> - hot_remove_disk(fd, makedev(disk.major, disk.minor));
> + hot_remove_disk(fd, makedev(disk.major, disk.minor), 1);
> }
> }
> c = map_num(pers, level);
> diff --git a/Manage.c b/Manage.c
> index edf5798aa87d..55218d9b0a7d 100644
> --- a/Manage.c
> +++ b/Manage.c
> @@ -1110,7 +1110,7 @@ int Manage_add(int fd, int tfd, struct mddev_dev *dv,
> }
>
> int Manage_remove(struct supertype *tst, int fd, struct mddev_dev *dv,
> - int sysfd, unsigned long rdev, int verbose, char *devname)
> + int sysfd, unsigned long rdev, int force, int verbose, char *devname)
> {
> int lfd = -1;
> int err;
> @@ -1177,9 +1177,9 @@ int Manage_remove(struct supertype *tst, int fd, struct mddev_dev *dv,
> /* device has been removed and we don't know
> * the major:minor number
> */
> - err = sys_hot_remove_disk(sysfd);
> + err = sys_hot_remove_disk(sysfd, force);
> } else {
> - err = hot_remove_disk(fd, rdev);
> + err = hot_remove_disk(fd, rdev, force);
> if (err && errno == ENODEV) {
> /* Old kernels rejected this if no personality
> * is registered */
> @@ -1603,7 +1603,7 @@ int Manage_subdevs(char *devname, int fd,
>
> if (dv->disposition == 'F')
> /* Need to remove first */
> - hot_remove_disk(fd, rdev);
> + hot_remove_disk(fd, rdev, force);
> /* Make sure it isn't in use (in 2.6 or later) */
> tfd = dev_open(dv->devname, O_RDONLY|O_EXCL);
> if (tfd >= 0) {
> @@ -1645,7 +1645,7 @@ int Manage_subdevs(char *devname, int fd,
> rv = -1;
> } else
> rv = Manage_remove(tst, fd, dv, sysfd,
> - rdev, verbose,
> + rdev, verbose, force,
> devname);
> if (sysfd >= 0)
> close(sysfd);
> diff --git a/mdadm.h b/mdadm.h
> index 52952a800ace..3b3d24230a79 100644
> --- a/mdadm.h
> +++ b/mdadm.h
> @@ -1476,8 +1476,8 @@ extern int add_disk(int mdfd, struct supertype *st,
> struct mdinfo *sra, struct mdinfo *info);
> extern int remove_disk(int mdfd, struct supertype *st,
> struct mdinfo *sra, struct mdinfo *info);
> -extern int hot_remove_disk(int mdfd, unsigned long dev);
> -extern int sys_hot_remove_disk(int statefd);
> +extern int hot_remove_disk(int mdfd, unsigned long dev, int force);
> +extern int sys_hot_remove_disk(int statefd, int force);
> extern int set_array_info(int mdfd, struct supertype *st, struct mdinfo *info);
> unsigned long long min_recovery_start(struct mdinfo *array);
>
> diff --git a/util.c b/util.c
> index 43fd83e72ded..6fd582520f0d 100644
> --- a/util.c
> +++ b/util.c
> @@ -1795,15 +1795,15 @@ int remove_disk(int mdfd, struct supertype *st,
> return rv;
> }
>
> -int hot_remove_disk(int mdfd, unsigned long dev)
> +int hot_remove_disk(int mdfd, unsigned long dev, int force)
> {
> - int cnt = 5;
> + int cnt = force ? 500 : 5;
> int ret;
>
> /* HOT_REMOVE_DISK can fail with EBUSY if there are
> * outstanding IO requests to the device.
> * In this case, it can be helpful to wait a little while,
> - * up to half a second, for that IO to flush.
> + * up to 5 seconds if 'force' is set, or 50 msec if not.
> */
> while ((ret = ioctl(mdfd, HOT_REMOVE_DISK, dev)) == -1 &&
> errno == EBUSY &&
> @@ -1813,9 +1813,9 @@ int hot_remove_disk(int mdfd, unsigned long dev)
> return ret;
> }
>
> -int sys_hot_remove_disk(int statefd)
> +int sys_hot_remove_disk(int statefd, int force)
> {
> - int cnt = 5;
> + int cnt = force ? 500 : 5;
> int ret;
>
> while ((ret = write(statefd, "remove", 6)) == -1 &&
>
>
> --
Ah, you've found and fix it,
Reviewed-by: Jack Wang <jinpu.wang@xxxxxxxxxxxxxxxx>
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
