Re: [RFC PATCH v3] mm, pmem, xfs: Introduce MF_MEM_REMOVE for unbind

[Shiyang Ruan <ruansy.fnst@fujitsu.com>]

在 2022/6/23 0:49, Darrick J. Wong 写道:
> On Wed, Jun 15, 2022 at 08:54:00PM +0800, Shiyang Ruan wrote:
>> This patch is inspired by Dan's "mm, dax, pmem: Introduce
>> dev_pagemap_failure()"[1].  With the help of dax_holder and
>> ->notify_failure() mechanism, the pmem driver is able to ask filesystem
>> (or mapped device) on it to unmap all files in use and notify processes
>> who are using those files.
>>
>> Call trace:
>> trigger unbind
>>   -> unbind_store()
>>    -> ... (skip)
>>     -> devres_release_all()   # was pmem driver ->remove() in v1
>>      -> kill_dax()
>>       -> dax_holder_notify_failure(dax_dev, 0, U64_MAX, MF_MEM_REMOVE)
>>        -> xfs_dax_notify_failure()
>>
>> Introduce MF_MEM_REMOVE to let filesystem know this is a remove event.
>> So do not shutdown filesystem directly if something not supported, or if
>> failure range includes metadata area.  Make sure all files and processes
>> are handled correctly.
>>
>> [1]: https://lore.kernel.org/linux-mm/161604050314.1463742.14151665140035795571.stgit@dwillia2-desk3.amr.corp.intel.com/
>>
>> Signed-off-by: Shiyang Ruan <ruansy.fnst@fujitsu.com>
>>
>> ==
>> Changes since v2:
>>    1. Rebased on next-20220615
>>
>> Changes since v1:
>>    1. Drop the needless change of moving {kill,put}_dax()
>>    2. Rebased on '[PATCHSETS] v14 fsdax-rmap + v11 fsdax-reflink'[2]
>>
>> ---
>>   drivers/dax/super.c         | 2 +-
>>   fs/xfs/xfs_notify_failure.c | 6 +++++-
>>   include/linux/mm.h          | 1 +
>>   3 files changed, 7 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/dax/super.c b/drivers/dax/super.c
>> index 9b5e2a5eb0ae..d4bc83159d46 100644
>> --- a/drivers/dax/super.c
>> +++ b/drivers/dax/super.c
>> @@ -323,7 +323,7 @@ void kill_dax(struct dax_device *dax_dev)
>>   		return;
>>   
>>   	if (dax_dev->holder_data != NULL)
>> -		dax_holder_notify_failure(dax_dev, 0, U64_MAX, 0);
>> +		dax_holder_notify_failure(dax_dev, 0, U64_MAX, MF_MEM_REMOVE);
> 
> At the point we're initiating a MEM_REMOVE call, is the pmem already
> gone, or is it about to be gone?

It's about to be gone.

I found two cases:
   1. exec `unbind` by user, who wants to unplug the pmem
   2. handle failures during initialization

> 
>>   
>>   	clear_bit(DAXDEV_ALIVE, &dax_dev->flags);
>>   	synchronize_srcu(&dax_srcu);
>> diff --git a/fs/xfs/xfs_notify_failure.c b/fs/xfs/xfs_notify_failure.c
>> index aa8dc27c599c..91d3f05d4241 100644
>> --- a/fs/xfs/xfs_notify_failure.c
>> +++ b/fs/xfs/xfs_notify_failure.c
>> @@ -73,7 +73,9 @@ xfs_dax_failure_fn(
>>   	struct failure_info		*notify = data;
>>   	int				error = 0;
>>   
>> -	if (XFS_RMAP_NON_INODE_OWNER(rec->rm_owner) ||
>> +	/* Do not shutdown so early when device is to be removed */
>> +	if (!(notify->mf_flags & MF_MEM_REMOVE) ||
>> +	    XFS_RMAP_NON_INODE_OWNER(rec->rm_owner) ||
>>   	    (rec->rm_flags & (XFS_RMAP_ATTR_FORK | XFS_RMAP_BMBT_BLOCK))) {
>>   		xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_ONDISK);
>>   		return -EFSCORRUPTED;
>> @@ -182,6 +184,8 @@ xfs_dax_notify_failure(
>>   
>>   	if (mp->m_logdev_targp && mp->m_logdev_targp->bt_daxdev == dax_dev &&
>>   	    mp->m_logdev_targp != mp->m_ddev_targp) {
>> +		if (mf_flags & MF_MEM_REMOVE)
>> +			return -EOPNOTSUPP;
> 
> The reason I ask is that if the pmem is *about to be* but not yet
> removed from the system, shouldn't we at least try to flush all dirty
> files and the log to reduce data loss and minimize recovery time?

Yes, they should be flushed.  Will add it.


--
Thanks,
Ruan.

> 
> If it's already gone, then you might as well shut down immediately,
> unless there's a chance the pmem will come back(?)
> 
> --D
> 
>>   		xfs_err(mp, "ondisk log corrupt, shutting down fs!");
>>   		xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_ONDISK);
>>   		return -EFSCORRUPTED;
>> diff --git a/include/linux/mm.h b/include/linux/mm.h
>> index 623c2ee8330a..bbeb31883362 100644
>> --- a/include/linux/mm.h
>> +++ b/include/linux/mm.h
>> @@ -3249,6 +3249,7 @@ enum mf_flags {
>>   	MF_SOFT_OFFLINE = 1 << 3,
>>   	MF_UNPOISON = 1 << 4,
>>   	MF_NO_RETRY = 1 << 5,
>> +	MF_MEM_REMOVE = 1 << 6,
>>   };
>>   int mf_dax_kill_procs(struct address_space *mapping, pgoff_t index,
>>   		      unsigned long count, int mf_flags);
>> -- 
>> 2.36.1
>>
>>
>>