mbox series

[0/11] btrfs: seed fix null ptr, use only main device_list_mutex, and cleanups

Message ID cover.1598792561.git.anand.jain@oracle.com
Headers show
Series btrfs: seed fix null ptr, use only main device_list_mutex, and cleanups | expand

Message

Anand Jain Aug. 31, 2020, 1:38 a.m. UTC
The cover letter for the patchset below [1] doesn't seem to have reached the ML.
So here I am sending it again.

[1]
 https://patchwork.kernel.org/project/linux-btrfs/list/?series=340699

In this patch set:
Rebased on latest misc-next which contains the seed_list from Nicolas.
This patchset has been sent before as two sets, here it is been merged.

Patches 1/11 fix the null kobject being put, which is observed with kernel
compiled with memory poisoning.
Patches 2/11 and 3/11 are cleanups to maintain better flow in the
functions btrfs_sysfs_add_devices_dir() and btrfs_sysfs_remove_devices_dir().
Patches 4/11 and 5/11 drops the last two users of seed device_list_mutex,
so now they hold main filesystem device_list_mutex. 
Patches [6-10]/11 are cleanups.
Patches 11/11 block the replacement of seed device in a sprouted FS, we already
block replacing a seed device on a non-sprouted device.

This set has passed through fstests/volume.

---- original cover-letter ---------
In a sprouted file system, the seed's fs_devices are cloned and 
listed under the sprout's fs_devices. The fs_info::fs_devices
points to the sprout::fs_devices and various critical operations
like device-delete holds the top-level main device_list_mutex
which is sprout::fs_devices::device_list_mutex and _not_ the seed level
mutex such as sprout::fs_devices::seed::fs_devices::device_list_mutex.

Also all related readers (except for two threads- reada and init_devices_late)
hold the sprout::fs_devices::device_list_mutex too. And those two threads
which are missing to hold the correct lock are being fixed here.

I take the approach to fix the read end instead of fixing the writer end
which are not holding the seed level mutex because to keep things
simple and there isn't much benefit burning extra CPU cycles in going
through the lock/unlock process as we traverse through the
fs_devices::seed fs_devices (for example as in reada and init_devices_late
threads).

The first two patches (1/7, 2/7) fixes the threads using the
seed::device_list_mutex.

And rest of the patches ([3-7]/7) are cleanups and these patches
are independent by themself.

These patchset has been tested with full xfstests btrfs test cases and
found to have no new regressions.


Anand Jain (11):
  btrfs: initialize sysfs devid and device link for seed device
  btrfs: refactor btrfs_sysfs_add_devices_dir
  btrfs: refactor btrfs_sysfs_remove_devices_dir
  btrfs: reada: use sprout device_list_mutex
  btrfs: btrfs_init_devices_late: use sprout device_list_mutex
  btrfs: open code list_head pointer in btrfs_init_dev_replace_tgtdev
  btrfs: cleanup btrfs_remove_chunk
  btrfs: cleanup btrfs_assign_next_active_device()
  btrfs: cleanup unnecessary goto in open_seed_device
  btrfs: btrfs_dev_replace_update_device_in_mapping_tree drop file
    global declare
  btrfs: fix replace of seed device

 fs/btrfs/dev-replace.c |  66 ++++++++----------
 fs/btrfs/reada.c       |   5 +-
 fs/btrfs/sysfs.c       | 151 ++++++++++++++++++++++++-----------------
 fs/btrfs/sysfs.h       |   8 +--
 fs/btrfs/volumes.c     |  40 +++++------
 5 files changed, 143 insertions(+), 127 deletions(-)