Skip to contentSkip to conversation

fscache,cachefiles: add prepare_ondemand_read() interface

PATCH In review Jingbo Xu 4 participants 6 messages Nov 24, 2022 – Nov 25, 2022

Subsystems (5)
lore
Share & tools
lore

Jingbo Xu

author
#0 lore raw

v5 cover fscache,cachefiles: add prepare_ondemand_read() interface

v5:
- patch 1: add back netfs_inode number to trace_cachefiles_prep_read, and .prepare_ondemand_read() now also accepts a "ino_t ino" parameter (David)
- add reviewed-by tags

v4: https://lore.kernel.org/all/[email protected]/
- patch 1
- make cachefiles_do_prepare_read() pass start by value (Jeff Layton)
- adjust the indentation of the parameter/argument list, so that they are all lined up (David)
- pass flags in for cachefiles_prepare_ondemand_read(), so that it can tail call cachefiles_do_prepare_read() directly without shuffling arguments around (David)
- declare cachefiles_do_prepare_read() as inline, to eliminate one extra function calling and arguments copying when calling cachefiles_do_prepare_read() (David)

v3:
- rebase to v6.1-rc5, while the xas_retry() checking in patch 2 has been extracted out as a separate fix [1]

[1] commit 37020bbb71d9 ("erofs: fix missing xas_retry() in fscache mode")
(https://github.com/torvalds/linux/commit/37020bbb71d9)

v2:
- patch 1: the generic routine, i.e. cachefiles_do_prepare_read() now accepts a parameter list instead of netfs_io_subrequest, and thus some debug info retrieved from netfs_io_subrequest is removed from trace_cachefiles_prep_read().
- patch 2: add xas_retry() checking in erofs_fscache_req_complete()

[Rationale]
===========
Fscache has been landed as a generic caching management framework in the Linux kernel for decades. It aims to manage cache data availability or fetch data if needed. Currently it's mainly used for network fses, but in principle the main caching subsystem can be used more widely.

We do really like fscache framework and we believe it'd be better to reuse such framework if possible instead of duplicating other alternatives for better maintenance and testing. Therefore for our container image use cases, we applied the existing fscache to implement on-demand read for erofs in the past months. For more details, also see [1].

In short, here each erofs filesystem is composed of multiple blobs (or devices). Each blob corresponds to one fscache cookie to strictly follow on-disk format and implement the image downloading in a deterministic manner, which means it has a unique checksum and is signed by vendors.

Data of each erofs inode can be scattered among multiple blobs (cookie) since erofs supports chunk-level deduplication. In this case, each erofs inode can correspond to multiple cookies, and there's a logical to physical offset mapping between the logical offset in erofs inode and the physical offset in the backing file.

As described above, per-cookie netfs model can not be used here directly. Instead, we'd like to propose/decouple a simple set of raw fscache APIs, to access cache for all fses to use. We believe it's useful since it's like the relationship between raw bio and iomap, both of which are useful for local fses. fscache_read() seems a reasonable candidate and is enough for such use case.

In addition, the on-demand read feature relies on .prepare_read() to reuse the hole detecting logic as much as possible. However, after fscache/netfs rework, libnetfs is preferred to access fscache, making .prepare_read() closely coupled with libnetfs, or more precisely, netfs_io_subrequest.

[What We Do]
============
As we discussed previously, we propose a new interface, i,e, .prepare_ondemand_read() dedicated for the on-demand read scenarios, which is independent on netfs_io_subrequest. The netfs will still use the original .prepare_read() as usual.

Jingbo Xu (2):
fscache,cachefiles: add prepare_ondemand_read() callback
erofs: switch to prepare_ondemand_read() in fscache mode

 fs/cachefiles/io.c                |  77 +++++----
 fs/erofs/fscache.c                | 261 +++++++++++-------------------
 include/linux/netfs.h             |   8 +
 include/trace/events/cachefiles.h |  27 ++--
 4 files changed, 166 insertions(+), 207 deletions(-)
Signature · 1 line
2.19.1.6.gb485710b
author
#1 lore raw

PATCH v5 1/2 fscache,cachefiles: add prepare_ondemand_read() callback

Add prepare_ondemand_read() callback dedicated for the on-demand read scenario, so that callers from this scenario can be decoupled from netfs_io_subrequest.

The original cachefiles_prepare_read() is now refactored to a generic routine accepting a parameter list instead of netfs_io_subrequest. There's no logic change, except that the debug id of subrequest and request is removed from trace_cachefiles_prep_read().

---
 fs/cachefiles/io.c                | 77 ++++++++++++++++++++-----------
 include/linux/netfs.h             |  8 ++++
 include/trace/events/cachefiles.h | 27 +++++------
 3 files changed, 72 insertions(+), 40 deletions(-)
fs/cachefiles/io.c +50 -27
index 000a28f46e59..175a25fcade8 100644
@@ -385,38 +385,35 @@ static int cachefiles_write(struct netfs_cache_resources *cres,
 				  term_func, term_func_priv);
 }
 
-/*
- * Prepare a read operation, shortening it to a cached/uncached
- * boundary as appropriate.
- */
-static enum netfs_io_source cachefiles_prepare_read(struct netfs_io_subrequest *subreq,
-						      loff_t i_size)
+static inline enum netfs_io_source
+cachefiles_do_prepare_read(struct netfs_cache_resources *cres,
+			   loff_t start, size_t *_len, loff_t i_size,
+			   unsigned long *_flags, ino_t netfs_ino)
 {
 	enum cachefiles_prepare_read_trace why;
-	struct netfs_io_request *rreq = subreq->rreq;
-	struct netfs_cache_resources *cres = &rreq->cache_resources;
-	struct cachefiles_object *object;
+	struct cachefiles_object *object = NULL;
 	struct cachefiles_cache *cache;
 	struct fscache_cookie *cookie = fscache_cres_cookie(cres);
 	const struct cred *saved_cred;
 	struct file *file = cachefiles_cres_file(cres);
 	enum netfs_io_source ret = NETFS_DOWNLOAD_FROM_SERVER;
+	size_t len = *_len;
 	loff_t off, to;
 	ino_t ino = file ? file_inode(file)->i_ino : 0;
 	int rc;
 
-	_enter("%zx @%llx/%llx", subreq->len, subreq->start, i_size);
+	_enter("%zx @%llx/%llx", len, start, i_size);
 
-	if (subreq->start >= i_size) {
+	if (start >= i_size) {
 		ret = NETFS_FILL_WITH_ZEROES;
 		why = cachefiles_trace_read_after_eof;
 		goto out_no_object;
 	}
 
 	if (test_bit(FSCACHE_COOKIE_NO_DATA_TO_READ, &cookie->flags)) {
-		__set_bit(NETFS_SREQ_COPY_TO_CACHE, &subreq->flags);
+		__set_bit(NETFS_SREQ_COPY_TO_CACHE, _flags);
 		why = cachefiles_trace_read_no_data;
-		if (!test_bit(NETFS_SREQ_ONDEMAND, &subreq->flags))
+		if (!test_bit(NETFS_SREQ_ONDEMAND, _flags))
 			goto out_no_object;
 	}
 
@@ -437,7 +434,7 @@ static enum netfs_io_source cachefiles_prepare_read(struct netfs_io_subrequest *
 retry:
 	off = cachefiles_inject_read_error();
 	if (off == 0)
-		off = vfs_llseek(file, subreq->start, SEEK_DATA);
+		off = vfs_llseek(file, start, SEEK_DATA);
 	if (off < 0 && off >= (loff_t)-MAX_ERRNO) {
 		if (off == (loff_t)-ENXIO) {
 			why = cachefiles_trace_read_seek_nxio;
@@ -449,21 +446,22 @@ static enum netfs_io_source cachefiles_prepare_read(struct netfs_io_subrequest *
 		goto out;
 	}
 
-	if (off >= subreq->start + subreq->len) {
+	if (off >= start + len) {
 		why = cachefiles_trace_read_found_hole;
 		goto download_and_store;
 	}
 
-	if (off > subreq->start) {
+	if (off > start) {
 		off = round_up(off, cache->bsize);
-		subreq->len = off - subreq->start;
+		len = off - start;
+		*_len = len;
 		why = cachefiles_trace_read_found_part;
 		goto download_and_store;
 	}
 
 	to = cachefiles_inject_read_error();
 	if (to == 0)
-		to = vfs_llseek(file, subreq->start, SEEK_HOLE);
+		to = vfs_llseek(file, start, SEEK_HOLE);
 	if (to < 0 && to >= (loff_t)-MAX_ERRNO) {
 		trace_cachefiles_io_error(object, file_inode(file), to,
 					  cachefiles_trace_seek_error);
@@ -471,12 +469,13 @@ static enum netfs_io_source cachefiles_prepare_read(struct netfs_io_subrequest *
 		goto out;
 	}
 
-	if (to < subreq->start + subreq->len) {
-		if (subreq->start + subreq->len >= i_size)
+	if (to < start + len) {
+		if (start + len >= i_size)
 			to = round_up(to, cache->bsize);
 		else
 			to = round_down(to, cache->bsize);
-		subreq->len = to - subreq->start;
+		len = to - start;
+		*_len = len;
 	}
 
 	why = cachefiles_trace_read_have_data;
@@ -484,12 +483,11 @@ static enum netfs_io_source cachefiles_prepare_read(struct netfs_io_subrequest *
 	goto out;
 
 download_and_store:
-	__set_bit(NETFS_SREQ_COPY_TO_CACHE, &subreq->flags);
-	if (test_bit(NETFS_SREQ_ONDEMAND, &subreq->flags)) {
-		rc = cachefiles_ondemand_read(object, subreq->start,
-					      subreq->len);
+	__set_bit(NETFS_SREQ_COPY_TO_CACHE, _flags);
+	if (test_bit(NETFS_SREQ_ONDEMAND, _flags)) {
+		rc = cachefiles_ondemand_read(object, start, len);
 		if (!rc) {
-			__clear_bit(NETFS_SREQ_ONDEMAND, &subreq->flags);
+			__clear_bit(NETFS_SREQ_ONDEMAND, _flags);
 			goto retry;
 		}
 		ret = NETFS_INVALID_READ;
@@ -497,10 +495,34 @@ static enum netfs_io_source cachefiles_prepare_read(struct netfs_io_subrequest *
 out:
 	cachefiles_end_secure(cache, saved_cred);
 out_no_object:
-	trace_cachefiles_prep_read(subreq, ret, why, ino);
+	trace_cachefiles_prep_read(object, start, len, *_flags, ret, why, ino, netfs_ino);
 	return ret;
 }
 
+/*
+ * Prepare a read operation, shortening it to a cached/uncached
+ * boundary as appropriate.
+ */
+static enum netfs_io_source cachefiles_prepare_read(struct netfs_io_subrequest *subreq,
+						    loff_t i_size)
+{
+	return cachefiles_do_prepare_read(&subreq->rreq->cache_resources,
+					  subreq->start, &subreq->len, i_size,
+					  &subreq->flags, subreq->rreq->inode->i_ino);
+}
+
+/*
+ * Prepare an on-demand read operation, shortening it to a cached/uncached
+ * boundary as appropriate.
+ */
+static enum netfs_io_source
+cachefiles_prepare_ondemand_read(struct netfs_cache_resources *cres,
+				 loff_t start, size_t *_len, loff_t i_size,
+				 unsigned long *_flags, ino_t ino)
+{
+	return cachefiles_do_prepare_read(cres, start, _len, i_size, _flags, ino);
+}
+
 /*
  * Prepare for a write to occur.
  */
@@ -621,6 +643,7 @@ static const struct netfs_cache_ops cachefiles_netfs_cache_ops = {
 	.write			= cachefiles_write,
 	.prepare_read		= cachefiles_prepare_read,
 	.prepare_write		= cachefiles_prepare_write,
+	.prepare_ondemand_read	= cachefiles_prepare_ondemand_read,
 	.query_occupancy	= cachefiles_query_occupancy,
 };
 
include/linux/netfs.h +8 -0
index f2402ddeafbf..4c76ddfb6a67 100644
@@ -267,6 +267,14 @@ struct netfs_cache_ops {
 			     loff_t *_start, size_t *_len, loff_t i_size,
 			     bool no_space_allocated_yet);
 
+	/* Prepare an on-demand read operation, shortening it to a cached/uncached
+	 * boundary as appropriate.
+	 */
+	enum netfs_io_source (*prepare_ondemand_read)(struct netfs_cache_resources *cres,
+						      loff_t start, size_t *_len,
+						      loff_t i_size,
+						      unsigned long *_flags, ino_t ino);
+
 	/* Query the occupancy of the cache in a region, returning where the
 	 * next chunk of data starts and how long it is.
 	 */
include/trace/events/cachefiles.h +14 -13
index d8d4d73fe7b6..cf4b98b9a9ed 100644
@@ -428,16 +428,18 @@ TRACE_EVENT(cachefiles_vol_coherency,
 	    );
 
 TRACE_EVENT(cachefiles_prep_read,
-	    TP_PROTO(struct netfs_io_subrequest *sreq,
+	    TP_PROTO(struct cachefiles_object *obj,
+		     loff_t start,
+		     size_t len,
+		     unsigned short flags,
 		     enum netfs_io_source source,
 		     enum cachefiles_prepare_read_trace why,
-		     ino_t cache_inode),
+		     ino_t cache_inode, ino_t netfs_inode),
 
-	    TP_ARGS(sreq, source, why, cache_inode),
+	    TP_ARGS(obj, start, len, flags, source, why, cache_inode, netfs_inode),
 
 	    TP_STRUCT__entry(
-		    __field(unsigned int,		rreq		)
-		    __field(unsigned short,		index		)
+		    __field(unsigned int,		obj		)
 		    __field(unsigned short,		flags		)
 		    __field(enum netfs_io_source,	source		)
 		    __field(enum cachefiles_prepare_read_trace,	why	)
@@ -448,19 +450,18 @@ TRACE_EVENT(cachefiles_prep_read,
 			     ),
 
 	    TP_fast_assign(
-		    __entry->rreq	= sreq->rreq->debug_id;
-		    __entry->index	= sreq->debug_index;
-		    __entry->flags	= sreq->flags;
+		    __entry->obj	= obj ? obj->debug_id : 0;
+		    __entry->flags	= flags;
 		    __entry->source	= source;
 		    __entry->why	= why;
-		    __entry->len	= sreq->len;
-		    __entry->start	= sreq->start;
-		    __entry->netfs_inode = sreq->rreq->inode->i_ino;
+		    __entry->len	= len;
+		    __entry->start	= start;
+		    __entry->netfs_inode = netfs_inode;
 		    __entry->cache_inode = cache_inode;
 			   ),
 
-	    TP_printk("R=%08x[%u] %s %s f=%02x s=%llx %zx ni=%x B=%x",
-		      __entry->rreq, __entry->index,
+	    TP_printk("o=%08x %s %s f=%02x s=%llx %zx ni=%x B=%x",
+		      __entry->obj,
 		      __print_symbolic(__entry->source, netfs_sreq_sources),
 		      __print_symbolic(__entry->why, cachefiles_prepare_read_traces),
 		      __entry->flags,
Signature · 1 line
2.19.1.6.gb485710b
maintainer Acked-by on v5 1/2
#2 lore raw

Re: [PATCH v5 1/2] fscache,cachefiles: add prepare_ondemand_read() callback

Quoted 11 lines · Jingbo Xu

Add prepare_ondemand_read() callback dedicated for the on-demand read scenario, so that callers from this scenario can be decoupled from netfs_io_subrequest.

The original cachefiles_prepare_read() is now refactored to a generic routine accepting a parameter list instead of netfs_io_subrequest. There's no logic change, except that the debug id of subrequest and request is removed from trace_cachefiles_prep_read().

maintainer
#3 lore raw

Re: [PATCH v5 1/2] fscache,cachefiles: add prepare_ondemand_read() callback

Hi David,

Quoted 11 lines · David Howells
Quoted 9 lines · Jingbo Xu

Add prepare_ondemand_read() callback dedicated for the on-demand read scenario, so that callers from this scenario can be decoupled from netfs_io_subrequest.

The original cachefiles_prepare_read() is now refactored to a generic routine accepting a parameter list instead of netfs_io_subrequest. There's no logic change, except that the debug id of subrequest and request is removed from trace_cachefiles_prep_read().

Quoted 4 lines
Quoted 2 lines

Thanks! I will apply them for -next, and soon Jingbo will submit large folios support for erofs iomap/fscache modes.

Thank all again,
Gao Xiang

author
#4 lore raw

PATCH v5 2/2 erofs: switch to prepare_ondemand_read() in fscache mode

In reply to Jingbo Xu

Switch to prepare_ondemand_read() interface and a self-contained request completion to get rid of netfs_io_[request|subrequest].

The whole request will still be split into slices (subrequest) according to the cache state of the backing file. As long as one of the subrequests fails, the whole request will be marked as failed.

---
 fs/erofs/fscache.c | 261 ++++++++++++++++-----------------------------
 1 file changed, 94 insertions(+), 167 deletions(-)
fs/erofs/fscache.c +94 -167
index af5ed6b9c54d..3cfe1af7a46e 100644
@@ -11,257 +11,180 @@ static DEFINE_MUTEX(erofs_domain_cookies_lock);
 static LIST_HEAD(erofs_domain_list);
 static struct vfsmount *erofs_pseudo_mnt;
 
-static struct netfs_io_request *erofs_fscache_alloc_request(struct address_space *mapping,
+struct erofs_fscache_request {
+	struct netfs_cache_resources cache_resources;
+	struct address_space	*mapping;	/* The mapping being accessed */
+	loff_t			start;		/* Start position */
+	size_t			len;		/* Length of the request */
+	size_t			submitted;	/* Length of submitted */
+	short			error;		/* 0 or error that occurred */
+	refcount_t		ref;
+};
+
+static struct erofs_fscache_request *erofs_fscache_req_alloc(struct address_space *mapping,
 					     loff_t start, size_t len)
 {
-	struct netfs_io_request *rreq;
+	struct erofs_fscache_request *req;
 
-	rreq = kzalloc(sizeof(struct netfs_io_request), GFP_KERNEL);
-	if (!rreq)
+	req = kzalloc(sizeof(struct erofs_fscache_request), GFP_KERNEL);
+	if (!req)
 		return ERR_PTR(-ENOMEM);
 
-	rreq->start	= start;
-	rreq->len	= len;
-	rreq->mapping	= mapping;
-	rreq->inode	= mapping->host;
-	INIT_LIST_HEAD(&rreq->subrequests);
-	refcount_set(&rreq->ref, 1);
-	return rreq;
-}
-
-static void erofs_fscache_put_request(struct netfs_io_request *rreq)
-{
-	if (!refcount_dec_and_test(&rreq->ref))
-		return;
-	if (rreq->cache_resources.ops)
-		rreq->cache_resources.ops->end_operation(&rreq->cache_resources);
-	kfree(rreq);
-}
-
-static void erofs_fscache_put_subrequest(struct netfs_io_subrequest *subreq)
-{
-	if (!refcount_dec_and_test(&subreq->ref))
-		return;
-	erofs_fscache_put_request(subreq->rreq);
-	kfree(subreq);
-}
+	req->mapping = mapping;
+	req->start   = start;
+	req->len     = len;
+	refcount_set(&req->ref, 1);
 
-static void erofs_fscache_clear_subrequests(struct netfs_io_request *rreq)
-{
-	struct netfs_io_subrequest *subreq;
-
-	while (!list_empty(&rreq->subrequests)) {
-		subreq = list_first_entry(&rreq->subrequests,
-				struct netfs_io_subrequest, rreq_link);
-		list_del(&subreq->rreq_link);
-		erofs_fscache_put_subrequest(subreq);
-	}
+	return req;
 }
 
-static void erofs_fscache_rreq_unlock_folios(struct netfs_io_request *rreq)
+static void erofs_fscache_req_complete(struct erofs_fscache_request *req)
 {
-	struct netfs_io_subrequest *subreq;
 	struct folio *folio;
-	unsigned int iopos = 0;
-	pgoff_t start_page = rreq->start / PAGE_SIZE;
-	pgoff_t last_page = ((rreq->start + rreq->len) / PAGE_SIZE) - 1;
-	bool subreq_failed = false;
+	bool failed = req->error;
+	pgoff_t start_page = req->start / PAGE_SIZE;
+	pgoff_t last_page = ((req->start + req->len) / PAGE_SIZE) - 1;
 
-	XA_STATE(xas, &rreq->mapping->i_pages, start_page);
-
-	subreq = list_first_entry(&rreq->subrequests,
-				  struct netfs_io_subrequest, rreq_link);
-	subreq_failed = (subreq->error < 0);
+	XA_STATE(xas, &req->mapping->i_pages, start_page);
 
 	rcu_read_lock();
 	xas_for_each(&xas, folio, last_page) {
-		unsigned int pgpos, pgend;
-		bool pg_failed = false;
-
 		if (xas_retry(&xas, folio))
 			continue;
-
-		pgpos = (folio_index(folio) - start_page) * PAGE_SIZE;
-		pgend = pgpos + folio_size(folio);
-
-		for (;;) {
-			if (!subreq) {
-				pg_failed = true;
-				break;
-			}
-
-			pg_failed |= subreq_failed;
-			if (pgend < iopos + subreq->len)
-				break;
-
-			iopos += subreq->len;
-			if (!list_is_last(&subreq->rreq_link,
-					  &rreq->subrequests)) {
-				subreq = list_next_entry(subreq, rreq_link);
-				subreq_failed = (subreq->error < 0);
-			} else {
-				subreq = NULL;
-				subreq_failed = false;
-			}
-			if (pgend == iopos)
-				break;
-		}
-
-		if (!pg_failed)
+		if (!failed)
 			folio_mark_uptodate(folio);
-
 		folio_unlock(folio);
 	}
 	rcu_read_unlock();
+
+	if (req->cache_resources.ops)
+		req->cache_resources.ops->end_operation(&req->cache_resources);
+
+	kfree(req);
 }
 
-static void erofs_fscache_rreq_complete(struct netfs_io_request *rreq)
+static void erofs_fscache_req_put(struct erofs_fscache_request *req)
 {
-	erofs_fscache_rreq_unlock_folios(rreq);
-	erofs_fscache_clear_subrequests(rreq);
-	erofs_fscache_put_request(rreq);
+	if (refcount_dec_and_test(&req->ref))
+		erofs_fscache_req_complete(req);
 }
 
-static void erofc_fscache_subreq_complete(void *priv,
+static void erofs_fscache_subreq_complete(void *priv,
 		ssize_t transferred_or_error, bool was_async)
 {
-	struct netfs_io_subrequest *subreq = priv;
-	struct netfs_io_request *rreq = subreq->rreq;
+	struct erofs_fscache_request *req = priv;
 
 	if (IS_ERR_VALUE(transferred_or_error))
-		subreq->error = transferred_or_error;
-
-	if (atomic_dec_and_test(&rreq->nr_outstanding))
-		erofs_fscache_rreq_complete(rreq);
-
-	erofs_fscache_put_subrequest(subreq);
+		req->error = transferred_or_error;
+	erofs_fscache_req_put(req);
 }
 
 /*
- * Read data from fscache and fill the read data into page cache described by
- * @rreq, which shall be both aligned with PAGE_SIZE. @pstart describes
- * the start physical address in the cache file.
+ * Read data from fscache (cookie, pstart, len), and fill the read data into
+ * page cache described by (req->mapping, lstart, len). @pstart describeis the
+ * start physical address in the cache file.
  */
 static int erofs_fscache_read_folios_async(struct fscache_cookie *cookie,
-				struct netfs_io_request *rreq, loff_t pstart)
+		struct erofs_fscache_request *req, loff_t pstart, size_t len)
 {
 	enum netfs_io_source source;
-	struct super_block *sb = rreq->mapping->host->i_sb;
-	struct netfs_io_subrequest *subreq;
-	struct netfs_cache_resources *cres = &rreq->cache_resources;
+	struct super_block *sb = req->mapping->host->i_sb;
+	struct netfs_cache_resources *cres = &req->cache_resources;
 	struct iov_iter iter;
-	loff_t start = rreq->start;
-	size_t len = rreq->len;
+	loff_t lstart = req->start + req->submitted;
 	size_t done = 0;
 	int ret;
 
-	atomic_set(&rreq->nr_outstanding, 1);
+	DBG_BUGON(len > req->len - req->submitted);
 
 	ret = fscache_begin_read_operation(cres, cookie);
 	if (ret)
-		goto out;
+		return ret;
 
 	while (done < len) {
-		subreq = kzalloc(sizeof(struct netfs_io_subrequest),
-				 GFP_KERNEL);
-		if (subreq) {
-			INIT_LIST_HEAD(&subreq->rreq_link);
-			refcount_set(&subreq->ref, 2);
-			subreq->rreq = rreq;
-			refcount_inc(&rreq->ref);
-		} else {
-			ret = -ENOMEM;
-			goto out;
-		}
-
-		subreq->start = pstart + done;
-		subreq->len	=  len - done;
-		subreq->flags = 1 << NETFS_SREQ_ONDEMAND;
+		loff_t sstart = pstart + done;
+		size_t slen = len - done;
+		unsigned long flags = 1 << NETFS_SREQ_ONDEMAND;
 
-		list_add_tail(&subreq->rreq_link, &rreq->subrequests);
-
-		source = cres->ops->prepare_read(subreq, LLONG_MAX);
-		if (WARN_ON(subreq->len == 0))
+		source = cres->ops->prepare_ondemand_read(cres,
+				sstart, &slen, LLONG_MAX, &flags, 0);
+		if (WARN_ON(slen == 0))
 			source = NETFS_INVALID_READ;
 		if (source != NETFS_READ_FROM_CACHE) {
-			erofs_err(sb, "failed to fscache prepare_read (source %d)",
-				  source);
-			ret = -EIO;
-			subreq->error = ret;
-			erofs_fscache_put_subrequest(subreq);
-			goto out;
+			erofs_err(sb, "failed to fscache prepare_read (source %d)", source);
+			return -EIO;
 		}
 
-		atomic_inc(&rreq->nr_outstanding);
+		refcount_inc(&req->ref);
+		iov_iter_xarray(&iter, READ, &req->mapping->i_pages,
+				lstart + done, slen);
 
-		iov_iter_xarray(&iter, READ, &rreq->mapping->i_pages,
-				start + done, subreq->len);
-
-		ret = fscache_read(cres, subreq->start, &iter,
-				   NETFS_READ_HOLE_FAIL,
-				   erofc_fscache_subreq_complete, subreq);
+		ret = fscache_read(cres, sstart, &iter, NETFS_READ_HOLE_FAIL,
+				   erofs_fscache_subreq_complete, req);
 		if (ret == -EIOCBQUEUED)
 			ret = 0;
 		if (ret) {
 			erofs_err(sb, "failed to fscache_read (ret %d)", ret);
-			goto out;
+			return ret;
 		}
 
-		done += subreq->len;
+		done += slen;
 	}
-out:
-	if (atomic_dec_and_test(&rreq->nr_outstanding))
-		erofs_fscache_rreq_complete(rreq);
-
-	return ret;
+	DBG_BUGON(done != len);
+	req->submitted += len;
+	return 0;
 }
 
 static int erofs_fscache_meta_read_folio(struct file *data, struct folio *folio)
 {
 	int ret;
 	struct super_block *sb = folio_mapping(folio)->host->i_sb;
-	struct netfs_io_request *rreq;
+	struct erofs_fscache_request *req;
 	struct erofs_map_dev mdev = {
 		.m_deviceid = 0,
 		.m_pa = folio_pos(folio),
 	};
 
 	ret = erofs_map_dev(sb, &mdev);
-	if (ret)
-		goto out;
+	if (ret) {
+		folio_unlock(folio);
+		return ret;
+	}
 
-	rreq = erofs_fscache_alloc_request(folio_mapping(folio),
+	req = erofs_fscache_req_alloc(folio_mapping(folio),
 				folio_pos(folio), folio_size(folio));
-	if (IS_ERR(rreq)) {
-		ret = PTR_ERR(rreq);
-		goto out;
+	if (IS_ERR(req)) {
+		folio_unlock(folio);
+		return PTR_ERR(req);
 	}
 
-	return erofs_fscache_read_folios_async(mdev.m_fscache->cookie,
-				rreq, mdev.m_pa);
-out:
-	folio_unlock(folio);
+	ret = erofs_fscache_read_folios_async(mdev.m_fscache->cookie,
+				req, mdev.m_pa, folio_size(folio));
+	if (ret)
+		req->error = ret;
+
+	erofs_fscache_req_put(req);
 	return ret;
 }
 
 /*
  * Read into page cache in the range described by (@pos, @len).
  *
- * On return, the caller is responsible for page unlocking if the output @unlock
- * is true, or the callee will take this responsibility through netfs_io_request
- * interface.
+ * On return, if the output @unlock is true, the caller is responsible for page
+ * unlocking; otherwise the callee will take this responsibility through request
+ * completion.
  *
  * The return value is the number of bytes successfully handled, or negative
  * error code on failure. The only exception is that, the length of the range
- * instead of the error code is returned on failure after netfs_io_request is
- * allocated, so that .readahead() could advance rac accordingly.
+ * instead of the error code is returned on failure after request is allocated,
+ * so that .readahead() could advance rac accordingly.
  */
 static int erofs_fscache_data_read(struct address_space *mapping,
 				   loff_t pos, size_t len, bool *unlock)
 {
 	struct inode *inode = mapping->host;
 	struct super_block *sb = inode->i_sb;
-	struct netfs_io_request *rreq;
+	struct erofs_fscache_request *req;
 	struct erofs_map_blocks map;
 	struct erofs_map_dev mdev;
 	struct iov_iter iter;
@@ -318,13 +241,17 @@ static int erofs_fscache_data_read(struct address_space *mapping,
 	if (ret)
 		return ret;
 
-	rreq = erofs_fscache_alloc_request(mapping, pos, count);
-	if (IS_ERR(rreq))
-		return PTR_ERR(rreq);
+	req = erofs_fscache_req_alloc(mapping, pos, count);
+	if (IS_ERR(req))
+		return PTR_ERR(req);
 
 	*unlock = false;
-	erofs_fscache_read_folios_async(mdev.m_fscache->cookie,
-			rreq, mdev.m_pa + (pos - map.m_la));
+	ret = erofs_fscache_read_folios_async(mdev.m_fscache->cookie,
+			req, mdev.m_pa + (pos - map.m_la), count);
+	if (ret)
+		req->error = ret;
+
+	erofs_fscache_req_put(req);
 	return count;
 }
 
Signature · 1 line
2.19.1.6.gb485710b
Reviewed-by on v5 2/2
#5 lore raw

Re: [Linux-cachefs] [PATCH v5 2/2] erofs: switch to prepare_ondemand_read() in fscache mode

Quoted 8 lines · Jingbo Xu

Switch to prepare_ondemand_read() interface and a self-contained request completion to get rid of netfs_io_[request|subrequest].

The whole request will still be split into slices (subrequest) according to the cache state of the backing file. As long as one of the subrequests fails, the whole request will be marked as failed.

Thanks.

Quoted 375 lines
---
  fs/erofs/fscache.c | 261 ++++++++++++++++-----------------------------
  1 file changed, 94 insertions(+), 167 deletions(-)
fs/erofs/fscache.c +94 -167
index af5ed6b9c54d..3cfe1af7a46e 100644
@@ -11,257 +11,180 @@ static DEFINE_MUTEX(erofs_domain_cookies_lock);
  static LIST_HEAD(erofs_domain_list);
  static struct vfsmount *erofs_pseudo_mnt;
 
-static struct netfs_io_request *erofs_fscache_alloc_request(struct address_space *mapping,
+struct erofs_fscache_request {
+	struct netfs_cache_resources cache_resources;
+	struct address_space	*mapping;	/* The mapping being accessed */
+	loff_t			start;		/* Start position */
+	size_t			len;		/* Length of the request */
+	size_t			submitted;	/* Length of submitted */
+	short			error;		/* 0 or error that occurred */
+	refcount_t		ref;
+};
+
+static struct erofs_fscache_request *erofs_fscache_req_alloc(struct address_space *mapping,
  					     loff_t start, size_t len)
  {
-	struct netfs_io_request *rreq;
+	struct erofs_fscache_request *req;
 
-	rreq = kzalloc(sizeof(struct netfs_io_request), GFP_KERNEL);
-	if (!rreq)
+	req = kzalloc(sizeof(struct erofs_fscache_request), GFP_KERNEL);
+	if (!req)
  		return ERR_PTR(-ENOMEM);
 
-	rreq->start	= start;
-	rreq->len	= len;
-	rreq->mapping	= mapping;
-	rreq->inode	= mapping->host;
-	INIT_LIST_HEAD(&rreq->subrequests);
-	refcount_set(&rreq->ref, 1);
-	return rreq;
-}
-
-static void erofs_fscache_put_request(struct netfs_io_request *rreq)
-{
-	if (!refcount_dec_and_test(&rreq->ref))
-		return;
-	if (rreq->cache_resources.ops)
-		rreq->cache_resources.ops->end_operation(&rreq->cache_resources);
-	kfree(rreq);
-}
-
-static void erofs_fscache_put_subrequest(struct netfs_io_subrequest *subreq)
-{
-	if (!refcount_dec_and_test(&subreq->ref))
-		return;
-	erofs_fscache_put_request(subreq->rreq);
-	kfree(subreq);
-}
+	req->mapping = mapping;
+	req->start   = start;
+	req->len     = len;
+	refcount_set(&req->ref, 1);
 
-static void erofs_fscache_clear_subrequests(struct netfs_io_request *rreq)
-{
-	struct netfs_io_subrequest *subreq;
-
-	while (!list_empty(&rreq->subrequests)) {
-		subreq = list_first_entry(&rreq->subrequests,
-				struct netfs_io_subrequest, rreq_link);
-		list_del(&subreq->rreq_link);
-		erofs_fscache_put_subrequest(subreq);
-	}
+	return req;
  }
 
-static void erofs_fscache_rreq_unlock_folios(struct netfs_io_request *rreq)
+static void erofs_fscache_req_complete(struct erofs_fscache_request *req)
  {
-	struct netfs_io_subrequest *subreq;
  	struct folio *folio;
-	unsigned int iopos = 0;
-	pgoff_t start_page = rreq->start / PAGE_SIZE;
-	pgoff_t last_page = ((rreq->start + rreq->len) / PAGE_SIZE) - 1;
-	bool subreq_failed = false;
+	bool failed = req->error;
+	pgoff_t start_page = req->start / PAGE_SIZE;
+	pgoff_t last_page = ((req->start + req->len) / PAGE_SIZE) - 1;
 
-	XA_STATE(xas, &rreq->mapping->i_pages, start_page);
-
-	subreq = list_first_entry(&rreq->subrequests,
-				  struct netfs_io_subrequest, rreq_link);
-	subreq_failed = (subreq->error < 0);
+	XA_STATE(xas, &req->mapping->i_pages, start_page);
 
  	rcu_read_lock();
  	xas_for_each(&xas, folio, last_page) {
-		unsigned int pgpos, pgend;
-		bool pg_failed = false;
-
  		if (xas_retry(&xas, folio))
  			continue;
-
-		pgpos = (folio_index(folio) - start_page) * PAGE_SIZE;
-		pgend = pgpos + folio_size(folio);
-
-		for (;;) {
-			if (!subreq) {
-				pg_failed = true;
-				break;
-			}
-
-			pg_failed |= subreq_failed;
-			if (pgend < iopos + subreq->len)
-				break;
-
-			iopos += subreq->len;
-			if (!list_is_last(&subreq->rreq_link,
-					  &rreq->subrequests)) {
-				subreq = list_next_entry(subreq, rreq_link);
-				subreq_failed = (subreq->error < 0);
-			} else {
-				subreq = NULL;
-				subreq_failed = false;
-			}
-			if (pgend == iopos)
-				break;
-		}
-
-		if (!pg_failed)
+		if (!failed)
  			folio_mark_uptodate(folio);
-
  		folio_unlock(folio);
  	}
  	rcu_read_unlock();
+
+	if (req->cache_resources.ops)
+		req->cache_resources.ops->end_operation(&req->cache_resources);
+
+	kfree(req);
  }
 
-static void erofs_fscache_rreq_complete(struct netfs_io_request *rreq)
+static void erofs_fscache_req_put(struct erofs_fscache_request *req)
  {
-	erofs_fscache_rreq_unlock_folios(rreq);
-	erofs_fscache_clear_subrequests(rreq);
-	erofs_fscache_put_request(rreq);
+	if (refcount_dec_and_test(&req->ref))
+		erofs_fscache_req_complete(req);
  }
 
-static void erofc_fscache_subreq_complete(void *priv,
+static void erofs_fscache_subreq_complete(void *priv,
  		ssize_t transferred_or_error, bool was_async)
  {
-	struct netfs_io_subrequest *subreq = priv;
-	struct netfs_io_request *rreq = subreq->rreq;
+	struct erofs_fscache_request *req = priv;
 
  	if (IS_ERR_VALUE(transferred_or_error))
-		subreq->error = transferred_or_error;
-
-	if (atomic_dec_and_test(&rreq->nr_outstanding))
-		erofs_fscache_rreq_complete(rreq);
-
-	erofs_fscache_put_subrequest(subreq);
+		req->error = transferred_or_error;
+	erofs_fscache_req_put(req);
  }
 
  /*
- * Read data from fscache and fill the read data into page cache described by
- * @rreq, which shall be both aligned with PAGE_SIZE. @pstart describes
- * the start physical address in the cache file.
+ * Read data from fscache (cookie, pstart, len), and fill the read data into
+ * page cache described by (req->mapping, lstart, len). @pstart describeis the
+ * start physical address in the cache file.
   */
  static int erofs_fscache_read_folios_async(struct fscache_cookie *cookie,
-				struct netfs_io_request *rreq, loff_t pstart)
+		struct erofs_fscache_request *req, loff_t pstart, size_t len)
  {
  	enum netfs_io_source source;
-	struct super_block *sb = rreq->mapping->host->i_sb;
-	struct netfs_io_subrequest *subreq;
-	struct netfs_cache_resources *cres = &rreq->cache_resources;
+	struct super_block *sb = req->mapping->host->i_sb;
+	struct netfs_cache_resources *cres = &req->cache_resources;
  	struct iov_iter iter;
-	loff_t start = rreq->start;
-	size_t len = rreq->len;
+	loff_t lstart = req->start + req->submitted;
  	size_t done = 0;
  	int ret;
 
-	atomic_set(&rreq->nr_outstanding, 1);
+	DBG_BUGON(len > req->len - req->submitted);
 
  	ret = fscache_begin_read_operation(cres, cookie);
  	if (ret)
-		goto out;
+		return ret;
 
  	while (done < len) {
-		subreq = kzalloc(sizeof(struct netfs_io_subrequest),
-				 GFP_KERNEL);
-		if (subreq) {
-			INIT_LIST_HEAD(&subreq->rreq_link);
-			refcount_set(&subreq->ref, 2);
-			subreq->rreq = rreq;
-			refcount_inc(&rreq->ref);
-		} else {
-			ret = -ENOMEM;
-			goto out;
-		}
-
-		subreq->start = pstart + done;
-		subreq->len	=  len - done;
-		subreq->flags = 1 << NETFS_SREQ_ONDEMAND;
+		loff_t sstart = pstart + done;
+		size_t slen = len - done;
+		unsigned long flags = 1 << NETFS_SREQ_ONDEMAND;
 
-		list_add_tail(&subreq->rreq_link, &rreq->subrequests);
-
-		source = cres->ops->prepare_read(subreq, LLONG_MAX);
-		if (WARN_ON(subreq->len == 0))
+		source = cres->ops->prepare_ondemand_read(cres,
+				sstart, &slen, LLONG_MAX, &flags, 0);
+		if (WARN_ON(slen == 0))
  			source = NETFS_INVALID_READ;
  		if (source != NETFS_READ_FROM_CACHE) {
-			erofs_err(sb, "failed to fscache prepare_read (source %d)",
-				  source);
-			ret = -EIO;
-			subreq->error = ret;
-			erofs_fscache_put_subrequest(subreq);
-			goto out;
+			erofs_err(sb, "failed to fscache prepare_read (source %d)", source);
+			return -EIO;
  		}
 
-		atomic_inc(&rreq->nr_outstanding);
+		refcount_inc(&req->ref);
+		iov_iter_xarray(&iter, READ, &req->mapping->i_pages,
+				lstart + done, slen);
 
-		iov_iter_xarray(&iter, READ, &rreq->mapping->i_pages,
-				start + done, subreq->len);
-
-		ret = fscache_read(cres, subreq->start, &iter,
-				   NETFS_READ_HOLE_FAIL,
-				   erofc_fscache_subreq_complete, subreq);
+		ret = fscache_read(cres, sstart, &iter, NETFS_READ_HOLE_FAIL,
+				   erofs_fscache_subreq_complete, req);
  		if (ret == -EIOCBQUEUED)
  			ret = 0;
  		if (ret) {
  			erofs_err(sb, "failed to fscache_read (ret %d)", ret);
-			goto out;
+			return ret;
  		}
 
-		done += subreq->len;
+		done += slen;
  	}
-out:
-	if (atomic_dec_and_test(&rreq->nr_outstanding))
-		erofs_fscache_rreq_complete(rreq);
-
-	return ret;
+	DBG_BUGON(done != len);
+	req->submitted += len;
+	return 0;
  }
 
  static int erofs_fscache_meta_read_folio(struct file *data, struct folio *folio)
  {
  	int ret;
  	struct super_block *sb = folio_mapping(folio)->host->i_sb;
-	struct netfs_io_request *rreq;
+	struct erofs_fscache_request *req;
  	struct erofs_map_dev mdev = {
  		.m_deviceid = 0,
  		.m_pa = folio_pos(folio),
  	};
 
  	ret = erofs_map_dev(sb, &mdev);
-	if (ret)
-		goto out;
+	if (ret) {
+		folio_unlock(folio);
+		return ret;
+	}
 
-	rreq = erofs_fscache_alloc_request(folio_mapping(folio),
+	req = erofs_fscache_req_alloc(folio_mapping(folio),
  				folio_pos(folio), folio_size(folio));
-	if (IS_ERR(rreq)) {
-		ret = PTR_ERR(rreq);
-		goto out;
+	if (IS_ERR(req)) {
+		folio_unlock(folio);
+		return PTR_ERR(req);
  	}
 
-	return erofs_fscache_read_folios_async(mdev.m_fscache->cookie,
-				rreq, mdev.m_pa);
-out:
-	folio_unlock(folio);
+	ret = erofs_fscache_read_folios_async(mdev.m_fscache->cookie,
+				req, mdev.m_pa, folio_size(folio));
+	if (ret)
+		req->error = ret;
+
+	erofs_fscache_req_put(req);
  	return ret;
  }
 
  /*
   * Read into page cache in the range described by (@pos, @len).
   *
- * On return, the caller is responsible for page unlocking if the output @unlock
- * is true, or the callee will take this responsibility through netfs_io_request
- * interface.
+ * On return, if the output @unlock is true, the caller is responsible for page
+ * unlocking; otherwise the callee will take this responsibility through request
+ * completion.
   *
   * The return value is the number of bytes successfully handled, or negative
   * error code on failure. The only exception is that, the length of the range
- * instead of the error code is returned on failure after netfs_io_request is
- * allocated, so that .readahead() could advance rac accordingly.
+ * instead of the error code is returned on failure after request is allocated,
+ * so that .readahead() could advance rac accordingly.
   */
  static int erofs_fscache_data_read(struct address_space *mapping,
  				   loff_t pos, size_t len, bool *unlock)
  {
  	struct inode *inode = mapping->host;
  	struct super_block *sb = inode->i_sb;
-	struct netfs_io_request *rreq;
+	struct erofs_fscache_request *req;
  	struct erofs_map_blocks map;
  	struct erofs_map_dev mdev;
  	struct iov_iter iter;
@@ -318,13 +241,17 @@ static int erofs_fscache_data_read(struct address_space *mapping,
  	if (ret)
  		return ret;
 
-	rreq = erofs_fscache_alloc_request(mapping, pos, count);
-	if (IS_ERR(rreq))
-		return PTR_ERR(rreq);
+	req = erofs_fscache_req_alloc(mapping, pos, count);
+	if (IS_ERR(req))
+		return PTR_ERR(req);
 
  	*unlock = false;
-	erofs_fscache_read_folios_async(mdev.m_fscache->cookie,
-			rreq, mdev.m_pa + (pos - map.m_la));
+	ret = erofs_fscache_read_folios_async(mdev.m_fscache->cookie,
+			req, mdev.m_pa + (pos - map.m_la), count);
+	if (ret)
+		req->error = ret;
+
+	erofs_fscache_req_put(req);
  	return count;
  }