Commit 20bf195e authored by Linus Torvalds's avatar Linus Torvalds
Browse files

Merge tag 'ceph-for-5.12-rc1' of git://github.com/ceph/ceph-client 

Pull ceph updates from Ilya Dryomov:
 "With netfs helper library and fscache rework delayed, just a few cap
  handling improvements to avoid grabbing mmap_lock in some code paths
  and deal with capsnaps better and a mount option cleanup"

* tag 'ceph-for-5.12-rc1' of git://github.com/ceph/ceph-client:
  ceph: defer flushing the capsnap if the Fb is used
  libceph: remove osdtimeout option entirely
  libceph: deprecate [no]cephx_require_signatures options
  ceph: allow queueing cap/snap handling after putting cap references
  ceph: clean up inode work queueing
  ceph: fix flush_snap logic after putting caps
parents 9fe19046 558b4510

fs/ceph/addr.c

+1 −1
Original line number Diff line number Diff line
@@ -1662,7 +1662,7 @@ static vm_fault_t ceph_page_mkwrite(struct vm_fault *vmf) 


	dout("page_mkwrite %p %llu~%zd dropping cap refs on %s ret %x\n",

	     inode, off, len, ceph_cap_string(got), ret);

	ceph_put_cap_refs(ci, got);

	ceph_put_cap_refs_async(ci, got);

out_free:

	ceph_restore_sigs(&oldset);

	sb_end_pagefault(inode->i_sb);

fs/ceph/caps.c

+50 −20
Original line number Diff line number Diff line
@@ -3027,6 +3027,12 @@ static int ceph_try_drop_cap_snap(struct ceph_inode_info *ci, 
	return 0;

}



enum put_cap_refs_mode {

	PUT_CAP_REFS_SYNC = 0,

	PUT_CAP_REFS_NO_CHECK,

	PUT_CAP_REFS_ASYNC,

};



/*

 * Release cap refs.

 *
@@ -3037,10 +3043,11 @@ static int ceph_try_drop_cap_snap(struct ceph_inode_info *ci, 
 * cap_snap, and wake up any waiters.

 */

static void __ceph_put_cap_refs(struct ceph_inode_info *ci, int had,

				bool skip_checking_caps)

				enum put_cap_refs_mode mode)

{

	struct inode *inode = &ci->vfs_inode;

	int last = 0, put = 0, flushsnaps = 0, wake = 0;

	bool check_flushsnaps = false;



	spin_lock(&ci->i_ceph_lock);

	if (had & CEPH_CAP_PIN)
@@ -3057,26 +3064,17 @@ static void __ceph_put_cap_refs(struct ceph_inode_info *ci, int had, 
	if (had & CEPH_CAP_FILE_BUFFER) {

		if (--ci->i_wb_ref == 0) {

			last++;

			/* put the ref held by ceph_take_cap_refs() */

			put++;

			check_flushsnaps = true;

		}

		dout("put_cap_refs %p wb %d -> %d (?)\n",

		     inode, ci->i_wb_ref+1, ci->i_wb_ref);

	}

	if (had & CEPH_CAP_FILE_WR)

	if (had & CEPH_CAP_FILE_WR) {

		if (--ci->i_wr_ref == 0) {

			last++;

			if (__ceph_have_pending_cap_snap(ci)) {

				struct ceph_cap_snap *capsnap =

					list_last_entry(&ci->i_cap_snaps,

							struct ceph_cap_snap,

							ci_item);

				capsnap->writing = 0;

				if (ceph_try_drop_cap_snap(ci, capsnap))

					put++;

				else if (__ceph_finish_cap_snap(ci, capsnap))

					flushsnaps = 1;

				wake = 1;

			}

			check_flushsnaps = true;

			if (ci->i_wrbuffer_ref_head == 0 &&

			    ci->i_dirty_caps == 0 &&

			    ci->i_flushing_caps == 0) {
@@ -3088,15 +3086,42 @@ static void __ceph_put_cap_refs(struct ceph_inode_info *ci, int had, 
			if (!__ceph_is_any_real_caps(ci) && ci->i_snap_realm)

				drop_inode_snap_realm(ci);

		}

	}

	if (check_flushsnaps && __ceph_have_pending_cap_snap(ci)) {

		struct ceph_cap_snap *capsnap =

			list_last_entry(&ci->i_cap_snaps,

					struct ceph_cap_snap,

					ci_item);



		capsnap->writing = 0;

		if (ceph_try_drop_cap_snap(ci, capsnap))

			/* put the ref held by ceph_queue_cap_snap() */

			put++;

		else if (__ceph_finish_cap_snap(ci, capsnap))

			flushsnaps = 1;

		wake = 1;

	}

	spin_unlock(&ci->i_ceph_lock);



	dout("put_cap_refs %p had %s%s%s\n", inode, ceph_cap_string(had),

	     last ? " last" : "", put ? " put" : "");



	if (last && !skip_checking_caps)

	switch (mode) {

	case PUT_CAP_REFS_SYNC:

		if (last)

			ceph_check_caps(ci, 0, NULL);

		else if (flushsnaps)

			ceph_flush_snaps(ci, NULL);

		break;

	case PUT_CAP_REFS_ASYNC:

		if (last)

			ceph_queue_check_caps(inode);

		else if (flushsnaps)

			ceph_queue_flush_snaps(inode);

		break;

	default:

		break;

	}

	if (wake)

		wake_up_all(&ci->i_cap_wq);

	while (put-- > 0)
@@ -3105,12 +3130,17 @@ static void __ceph_put_cap_refs(struct ceph_inode_info *ci, int had, 


void ceph_put_cap_refs(struct ceph_inode_info *ci, int had)

{

	__ceph_put_cap_refs(ci, had, false);

	__ceph_put_cap_refs(ci, had, PUT_CAP_REFS_SYNC);

}



void ceph_put_cap_refs_async(struct ceph_inode_info *ci, int had)

{

	__ceph_put_cap_refs(ci, had, PUT_CAP_REFS_ASYNC);

}



void ceph_put_cap_refs_no_check_caps(struct ceph_inode_info *ci, int had)

{

	__ceph_put_cap_refs(ci, had, true);

	__ceph_put_cap_refs(ci, had, PUT_CAP_REFS_NO_CHECK);

}



/*

fs/ceph/inode.c

+12 −49
Original line number Diff line number Diff line
@@ -1816,60 +1816,17 @@ void ceph_async_iput(struct inode *inode) 
	}

}



/*

 * Write back inode data in a worker thread.  (This can't be done

 * in the message handler context.)

 */

void ceph_queue_writeback(struct inode *inode)

{

	struct ceph_inode_info *ci = ceph_inode(inode);

	set_bit(CEPH_I_WORK_WRITEBACK, &ci->i_work_mask);



	ihold(inode);

	if (queue_work(ceph_inode_to_client(inode)->inode_wq,

		       &ci->i_work)) {

		dout("ceph_queue_writeback %p\n", inode);

	} else {

		dout("ceph_queue_writeback %p already queued, mask=%lx\n",

		     inode, ci->i_work_mask);

		iput(inode);

	}

}



/*

 * queue an async invalidation

 */

void ceph_queue_invalidate(struct inode *inode)

{

	struct ceph_inode_info *ci = ceph_inode(inode);

	set_bit(CEPH_I_WORK_INVALIDATE_PAGES, &ci->i_work_mask);



	ihold(inode);

	if (queue_work(ceph_inode_to_client(inode)->inode_wq,

		       &ceph_inode(inode)->i_work)) {

		dout("ceph_queue_invalidate %p\n", inode);

	} else {

		dout("ceph_queue_invalidate %p already queued, mask=%lx\n",

		     inode, ci->i_work_mask);

		iput(inode);

	}

}



/*

 * Queue an async vmtruncate.  If we fail to queue work, we will handle

 * the truncation the next time we call __ceph_do_pending_vmtruncate.

 */

void ceph_queue_vmtruncate(struct inode *inode)

void ceph_queue_inode_work(struct inode *inode, int work_bit)

{

	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);

	struct ceph_inode_info *ci = ceph_inode(inode);

	set_bit(CEPH_I_WORK_VMTRUNCATE, &ci->i_work_mask);

	set_bit(work_bit, &ci->i_work_mask);



	ihold(inode);

	if (queue_work(ceph_inode_to_client(inode)->inode_wq,

		       &ci->i_work)) {

		dout("ceph_queue_vmtruncate %p\n", inode);

	if (queue_work(fsc->inode_wq, &ci->i_work)) {

		dout("queue_inode_work %p, mask=%lx\n", inode, ci->i_work_mask);

	} else {

		dout("ceph_queue_vmtruncate %p already queued, mask=%lx\n",

		dout("queue_inode_work %p already queued, mask=%lx\n",

		     inode, ci->i_work_mask);

		iput(inode);

	}
@@ -2008,6 +1965,12 @@ static void ceph_inode_work(struct work_struct *work) 
	if (test_and_clear_bit(CEPH_I_WORK_VMTRUNCATE, &ci->i_work_mask))

		__ceph_do_pending_vmtruncate(inode);



	if (test_and_clear_bit(CEPH_I_WORK_CHECK_CAPS, &ci->i_work_mask))

		ceph_check_caps(ci, 0, NULL);



	if (test_and_clear_bit(CEPH_I_WORK_FLUSH_SNAPS, &ci->i_work_mask))

		ceph_flush_snaps(ci, NULL);



	iput(inode);

}

fs/ceph/snap.c

+10 −0
Original line number Diff line number Diff line
@@ -623,6 +623,16 @@ int __ceph_finish_cap_snap(struct ceph_inode_info *ci, 
		return 0;

	}



	/* Fb cap still in use, delay it */

	if (ci->i_wb_ref) {

		dout("finish_cap_snap %p cap_snap %p snapc %p %llu %s s=%llu "

		     "used WRBUFFER, delaying\n", inode, capsnap,

		     capsnap->context, capsnap->context->seq,

		     ceph_cap_string(capsnap->dirty), capsnap->size);

		capsnap->writing = 1;

		return 0;

	}



	ci->i_ceph_flags |= CEPH_I_FLUSH_SNAPS;

	dout("finish_cap_snap %p cap_snap %p snapc %p %llu %s s=%llu\n",

	     inode, capsnap, capsnap->context,

fs/ceph/super.h

+34 −6
Original line number Diff line number Diff line
@@ -562,9 +562,11 @@ static inline struct inode *ceph_find_inode(struct super_block *sb, 
/*

 * Masks of ceph inode work.

 */

#define CEPH_I_WORK_WRITEBACK		0 /* writeback */

#define CEPH_I_WORK_INVALIDATE_PAGES	1 /* invalidate pages */

#define CEPH_I_WORK_VMTRUNCATE		2 /* vmtruncate */

#define CEPH_I_WORK_WRITEBACK		0

#define CEPH_I_WORK_INVALIDATE_PAGES	1

#define CEPH_I_WORK_VMTRUNCATE		2

#define CEPH_I_WORK_CHECK_CAPS		3

#define CEPH_I_WORK_FLUSH_SNAPS		4



/*

 * We set the ERROR_WRITE bit when we start seeing write errors on an inode
@@ -962,11 +964,36 @@ extern int ceph_inode_holds_cap(struct inode *inode, int mask); 


extern bool ceph_inode_set_size(struct inode *inode, loff_t size);

extern void __ceph_do_pending_vmtruncate(struct inode *inode);

extern void ceph_queue_vmtruncate(struct inode *inode);

extern void ceph_queue_invalidate(struct inode *inode);

extern void ceph_queue_writeback(struct inode *inode);



extern void ceph_async_iput(struct inode *inode);



void ceph_queue_inode_work(struct inode *inode, int work_bit);



static inline void ceph_queue_vmtruncate(struct inode *inode)

{

	ceph_queue_inode_work(inode, CEPH_I_WORK_VMTRUNCATE);

}



static inline void ceph_queue_invalidate(struct inode *inode)

{

	ceph_queue_inode_work(inode, CEPH_I_WORK_INVALIDATE_PAGES);

}



static inline void ceph_queue_writeback(struct inode *inode)

{

	ceph_queue_inode_work(inode, CEPH_I_WORK_WRITEBACK);

}



static inline void ceph_queue_check_caps(struct inode *inode)

{

	ceph_queue_inode_work(inode, CEPH_I_WORK_CHECK_CAPS);

}



static inline void ceph_queue_flush_snaps(struct inode *inode)

{

	ceph_queue_inode_work(inode, CEPH_I_WORK_FLUSH_SNAPS);

}



extern int __ceph_do_getattr(struct inode *inode, struct page *locked_page,

			     int mask, bool force);

static inline int ceph_do_getattr(struct inode *inode, int mask, bool force)
@@ -1105,6 +1132,7 @@ extern void ceph_take_cap_refs(struct ceph_inode_info *ci, int caps, 
				bool snap_rwsem_locked);

extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps);

extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had);

extern void ceph_put_cap_refs_async(struct ceph_inode_info *ci, int had);

extern void ceph_put_cap_refs_no_check_caps(struct ceph_inode_info *ci,

					    int had);

extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,