ANDROID: fs: FS tracepoints to track IO.

Adds tracepoints in ext4/f2fs/mpage to track readpages/buffered
write()s. This allows us to track files that are being read/written
to PIDs.

Bug: 120445624
Change-Id: I44476230324e9397e292328463f846af4befbd6d
[joelaf: Needed for storaged fsync accounting ("storaged --uid" and
         "storaged --task".)]
Signed-off-by: Mohan Srinivasan <srmohan@google.com>
[AmitP: Folded following android-4.9 commit changes into this patch
        a5c4dbb05ab7 ("ANDROID: Replace spaces by '_' for some
        android filesystem tracepoints.")]
Signed-off-by: Amit Pundir <amit.pundir@linaro.org>
[astrachan: Folded 63066f4acf92 ("ANDROID: fs: Refactor FS
            readpage/write tracepoints.") into this patch
Signed-off-by: Alistair Strachan <astrachan@google.com>

fs/ext4/inline.c

@@ -12,6 +12,7 @@
 #include "ext4.h"
 #include "xattr.h"
 #include "truncate.h"
+#include <trace/events/android_fs.h>
 
 #define EXT4_XATTR_SYSTEM_DATA	"data"
 #define EXT4_MIN_INLINE_DATA_SIZE	((sizeof(__le32) * EXT4_N_BLOCKS))
@@ -505,6 +506,17 @@ int ext4_readpage_inline(struct inode *inode, struct page *page)
 		return -EAGAIN;
 	}
 
+	if (trace_android_fs_dataread_start_enabled()) {
+		char *path, pathbuf[MAX_TRACE_PATHBUF_LEN];
+
+		path = android_fstrace_get_pathname(pathbuf,
+						    MAX_TRACE_PATHBUF_LEN,
+						    inode);
+		trace_android_fs_dataread_start(inode, page_offset(page),
+						PAGE_SIZE, current->pid,
+						path, current->comm);
+	}
+
 	/*
 	 * Current inline data can only exist in the 1st page,
 	 * So for all the other pages, just set them uptodate.
@@ -516,6 +528,8 @@ int ext4_readpage_inline(struct inode *inode, struct page *page)
 		SetPageUptodate(page);
 	}
 
+	trace_android_fs_dataread_end(inode, page_offset(page), PAGE_SIZE);
+
 	up_read(&EXT4_I(inode)->xattr_sem);
 
 	unlock_page(page);

fs/ext4/inode.c

@@ -47,6 +47,7 @@
 #include "truncate.h"
 
 #include <trace/events/ext4.h>
+#include <trace/events/android_fs.h>
 
 #define MPAGE_DA_EXTENT_TAIL 0x01
 
@@ -1253,6 +1254,16 @@ static int ext4_write_begin(struct file *file, struct address_space *mapping,
 	if (unlikely(ext4_forced_shutdown(EXT4_SB(inode->i_sb))))
 		return -EIO;
 
+	if (trace_android_fs_datawrite_start_enabled()) {
+		char *path, pathbuf[MAX_TRACE_PATHBUF_LEN];
+
+		path = android_fstrace_get_pathname(pathbuf,
+						    MAX_TRACE_PATHBUF_LEN,
+						    inode);
+		trace_android_fs_datawrite_start(inode, pos, len,
+						 current->pid, path,
+						 current->comm);
+	}
 	trace_ext4_write_begin(inode, pos, len, flags);
 	/*
 	 * Reserve one block more for addition to orphan list in case
@@ -1391,6 +1402,7 @@ static int ext4_write_end(struct file *file,
 	int i_size_changed = 0;
 	int inline_data = ext4_has_inline_data(inode);
 
+	trace_android_fs_datawrite_end(inode, pos, len);
 	trace_ext4_write_end(inode, pos, len, copied);
 	if (inline_data) {
 		ret = ext4_write_inline_data_end(inode, pos, len,
@@ -1496,6 +1508,7 @@ static int ext4_journalled_write_end(struct file *file,
 	int size_changed = 0;
 	int inline_data = ext4_has_inline_data(inode);
 
+	trace_android_fs_datawrite_end(inode, pos, len);
 	trace_ext4_journalled_write_end(inode, pos, len, copied);
 	from = pos & (PAGE_SIZE - 1);
 	to = from + len;
@@ -3024,6 +3037,16 @@ static int ext4_da_write_begin(struct file *file, struct address_space *mapping,
 					len, flags, pagep, fsdata);
 	}
 	*fsdata = (void *)0;
+	if (trace_android_fs_datawrite_start_enabled()) {
+		char *path, pathbuf[MAX_TRACE_PATHBUF_LEN];
+
+		path = android_fstrace_get_pathname(pathbuf,
+						    MAX_TRACE_PATHBUF_LEN,
+						    inode);
+		trace_android_fs_datawrite_start(inode, pos, len,
+						 current->pid,
+						 path, current->comm);
+	}
 	trace_ext4_da_write_begin(inode, pos, len, flags);
 
 	if (ext4_test_inode_state(inode, EXT4_STATE_MAY_INLINE_DATA)) {
@@ -3142,6 +3165,7 @@ static int ext4_da_write_end(struct file *file,
 		return ext4_write_end(file, mapping, pos,
 				      len, copied, page, fsdata);
 
+	trace_android_fs_datawrite_end(inode, pos, len);
 	trace_ext4_da_write_end(inode, pos, len, copied);
 	start = pos & (PAGE_SIZE - 1);
 	end = start + copied - 1;
@@ -3847,6 +3871,7 @@ static ssize_t ext4_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
 	size_t count = iov_iter_count(iter);
 	loff_t offset = iocb->ki_pos;
 	ssize_t ret;
+	int rw = iov_iter_rw(iter);
 
 #ifdef CONFIG_EXT4_FS_ENCRYPTION
 	if (ext4_encrypted_inode(inode) && S_ISREG(inode->i_mode))
@@ -3863,12 +3888,42 @@ static ssize_t ext4_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
 	if (ext4_has_inline_data(inode))
 		return 0;
 
+	if (trace_android_fs_dataread_start_enabled() &&
+	    (rw == READ)) {
+		char *path, pathbuf[MAX_TRACE_PATHBUF_LEN];
+
+		path = android_fstrace_get_pathname(pathbuf,
+						    MAX_TRACE_PATHBUF_LEN,
+						    inode);
+		trace_android_fs_dataread_start(inode, offset, count,
+						current->pid, path,
+						current->comm);
+	}
+	if (trace_android_fs_datawrite_start_enabled() &&
+	    (rw == WRITE)) {
+		char *path, pathbuf[MAX_TRACE_PATHBUF_LEN];
+
+		path = android_fstrace_get_pathname(pathbuf,
+						    MAX_TRACE_PATHBUF_LEN,
+						    inode);
+		trace_android_fs_datawrite_start(inode, offset, count,
+						 current->pid, path,
+						 current->comm);
+	}
 	trace_ext4_direct_IO_enter(inode, offset, count, iov_iter_rw(iter));
 	if (iov_iter_rw(iter) == READ)
 		ret = ext4_direct_IO_read(iocb, iter);
 	else
 		ret = ext4_direct_IO_write(iocb, iter);
 	trace_ext4_direct_IO_exit(inode, offset, count, iov_iter_rw(iter), ret);
+
+	if (trace_android_fs_dataread_start_enabled() &&
+	    (rw == READ))
+		trace_android_fs_dataread_end(inode, offset, count);
+	if (trace_android_fs_datawrite_start_enabled() &&
+	    (rw == WRITE))
+		trace_android_fs_datawrite_end(inode, offset, count);
+
 	return ret;
 }
 

fs/ext4/readpage.c

@@ -46,6 +46,7 @@
 #include <linux/cleancache.h>
 
 #include "ext4.h"
+#include <trace/events/android_fs.h>
 
 static inline bool ext4_bio_encrypted(struct bio *bio)
 {
@@ -56,6 +57,17 @@ static inline bool ext4_bio_encrypted(struct bio *bio)
 #endif
 }
 
+static void
+ext4_trace_read_completion(struct bio *bio)
+{
+	struct page *first_page = bio->bi_io_vec[0].bv_page;
+
+	if (first_page != NULL)
+		trace_android_fs_dataread_end(first_page->mapping->host,
+					      page_offset(first_page),
+					      bio->bi_iter.bi_size);
+}
+
 /*
  * I/O completion handler for multipage BIOs.
  *
@@ -73,6 +85,9 @@ static void mpage_end_io(struct bio *bio)
 	struct bio_vec *bv;
 	int i;
 
+	if (trace_android_fs_dataread_start_enabled())
+		ext4_trace_read_completion(bio);
+
 	if (ext4_bio_encrypted(bio)) {
 		if (bio->bi_status) {
 			fscrypt_release_ctx(bio->bi_private);
@@ -96,6 +111,30 @@ static void mpage_end_io(struct bio *bio)
 	bio_put(bio);
 }
 
+static void
+ext4_submit_bio_read(struct bio *bio)
+{
+	if (trace_android_fs_dataread_start_enabled()) {
+		struct page *first_page = bio->bi_io_vec[0].bv_page;
+
+		if (first_page != NULL) {
+			char *path, pathbuf[MAX_TRACE_PATHBUF_LEN];
+
+			path = android_fstrace_get_pathname(pathbuf,
+						    MAX_TRACE_PATHBUF_LEN,
+						    first_page->mapping->host);
+			trace_android_fs_dataread_start(
+				first_page->mapping->host,
+				page_offset(first_page),
+				bio->bi_iter.bi_size,
+				current->pid,
+				path,
+				current->comm);
+		}
+	}
+	submit_bio(bio);
+}
+
 int ext4_mpage_readpages(struct address_space *mapping,
 			 struct list_head *pages, struct page *page,
 			 unsigned nr_pages, bool is_readahead)
@@ -236,7 +275,7 @@ int ext4_mpage_readpages(struct address_space *mapping,
 		 */
 		if (bio && (last_block_in_bio != blocks[0] - 1)) {
 		submit_and_realloc:
-			submit_bio(bio);
+			ext4_submit_bio_read(bio);
 			bio = NULL;
 		}
 		if (bio == NULL) {
@@ -270,14 +309,14 @@ int ext4_mpage_readpages(struct address_space *mapping,
 		if (((map.m_flags & EXT4_MAP_BOUNDARY) &&
 		     (relative_block == map.m_len)) ||
 		    (first_hole != blocks_per_page)) {
-			submit_bio(bio);
+			ext4_submit_bio_read(bio);
 			bio = NULL;
 		} else
 			last_block_in_bio = blocks[blocks_per_page - 1];
 		goto next_page;
 	confused:
 		if (bio) {
-			submit_bio(bio);
+			ext4_submit_bio_read(bio);
 			bio = NULL;
 		}
 		if (!PageUptodate(page))
@@ -290,6 +329,6 @@ int ext4_mpage_readpages(struct address_space *mapping,
 	}
 	BUG_ON(pages && !list_empty(pages));
 	if (bio)
-		submit_bio(bio);
+		ext4_submit_bio_read(bio);
 	return 0;
 }

fs/f2fs/data.c

@@ -27,6 +27,7 @@
 #include "segment.h"
 #include "trace.h"
 #include <trace/events/f2fs.h>
+#include <trace/events/android_fs.h>
 
 #define NUM_PREALLOC_POST_READ_CTXS	128
 
@@ -2337,6 +2338,16 @@ static int f2fs_write_begin(struct file *file, struct address_space *mapping,
 	block_t blkaddr = NULL_ADDR;
 	int err = 0;
 
+	if (trace_android_fs_datawrite_start_enabled()) {
+		char *path, pathbuf[MAX_TRACE_PATHBUF_LEN];
+
+		path = android_fstrace_get_pathname(pathbuf,
+						    MAX_TRACE_PATHBUF_LEN,
+						    inode);
+		trace_android_fs_datawrite_start(inode, pos, len,
+						 current->pid, path,
+						 current->comm);
+	}
 	trace_f2fs_write_begin(inode, pos, len, flags);
 
 	if ((f2fs_is_atomic_file(inode) &&
@@ -2436,6 +2447,7 @@ static int f2fs_write_end(struct file *file,
 {
 	struct inode *inode = page->mapping->host;
 
+	trace_android_fs_datawrite_end(inode, pos, len);
 	trace_f2fs_write_end(inode, pos, len, copied);
 
 	/*
@@ -2503,6 +2515,29 @@ static ssize_t f2fs_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
 
 	trace_f2fs_direct_IO_enter(inode, offset, count, rw);
 
+	if (trace_android_fs_dataread_start_enabled() &&
+	    (rw == READ)) {
+		char *path, pathbuf[MAX_TRACE_PATHBUF_LEN];
+
+		path = android_fstrace_get_pathname(pathbuf,
+						    MAX_TRACE_PATHBUF_LEN,
+						    inode);
+		trace_android_fs_dataread_start(inode, offset,
+						count, current->pid, path,
+						current->comm);
+	}
+	if (trace_android_fs_datawrite_start_enabled() &&
+	    (rw == WRITE)) {
+		char *path, pathbuf[MAX_TRACE_PATHBUF_LEN];
+
+		path = android_fstrace_get_pathname(pathbuf,
+						    MAX_TRACE_PATHBUF_LEN,
+						    inode);
+		trace_android_fs_datawrite_start(inode, offset, count,
+						 current->pid, path,
+						 current->comm);
+	}
+
 	if (rw == WRITE && whint_mode == WHINT_MODE_OFF)
 		iocb->ki_hint = WRITE_LIFE_NOT_SET;
 
@@ -2531,6 +2566,13 @@ static ssize_t f2fs_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
 	}
 
 out:
+	if (trace_android_fs_dataread_start_enabled() &&
+	    (rw == READ))
+		trace_android_fs_dataread_end(inode, offset, count);
+	if (trace_android_fs_datawrite_start_enabled() &&
+	    (rw == WRITE))
+		trace_android_fs_datawrite_end(inode, offset, count);
+
 	trace_f2fs_direct_IO_exit(inode, offset, count, rw, err);
 
 	return err;

fs/f2fs/inline.c

@@ -13,6 +13,7 @@
 
 #include "f2fs.h"
 #include "node.h"
+#include <trace/events/android_fs.h>
 
 bool f2fs_may_inline_data(struct inode *inode)
 {
@@ -86,14 +87,29 @@ int f2fs_read_inline_data(struct inode *inode, struct page *page)
 {
 	struct page *ipage;
 
+	if (trace_android_fs_dataread_start_enabled()) {
+		char *path, pathbuf[MAX_TRACE_PATHBUF_LEN];
+
+		path = android_fstrace_get_pathname(pathbuf,
+						    MAX_TRACE_PATHBUF_LEN,
+						    inode);
+		trace_android_fs_dataread_start(inode, page_offset(page),
+						PAGE_SIZE, current->pid,
+						path, current->comm);
+	}
+
 	ipage = f2fs_get_node_page(F2FS_I_SB(inode), inode->i_ino);
 	if (IS_ERR(ipage)) {
+		trace_android_fs_dataread_end(inode, page_offset(page),
+					      PAGE_SIZE);
 		unlock_page(page);
 		return PTR_ERR(ipage);
 	}
 
 	if (!f2fs_has_inline_data(inode)) {
 		f2fs_put_page(ipage, 1);
+		trace_android_fs_dataread_end(inode, page_offset(page),
+					      PAGE_SIZE);
 		return -EAGAIN;
 	}
 
@@ -105,6 +121,8 @@ int f2fs_read_inline_data(struct inode *inode, struct page *page)
 	if (!PageUptodate(page))
 		SetPageUptodate(page);
 	f2fs_put_page(ipage, 1);
+	trace_android_fs_dataread_end(inode, page_offset(page),
+				      PAGE_SIZE);
 	unlock_page(page);
 	return 0;
 }

fs/mpage.c

@@ -32,6 +32,14 @@
 #include <linux/cleancache.h>
 #include "internal.h"
 
+#define CREATE_TRACE_POINTS
+#include <trace/events/android_fs.h>
+
+EXPORT_TRACEPOINT_SYMBOL(android_fs_datawrite_start);
+EXPORT_TRACEPOINT_SYMBOL(android_fs_datawrite_end);
+EXPORT_TRACEPOINT_SYMBOL(android_fs_dataread_start);
+EXPORT_TRACEPOINT_SYMBOL(android_fs_dataread_end);
+
 /*
  * I/O completion handler for multipage BIOs.
  *
@@ -49,6 +57,16 @@ static void mpage_end_io(struct bio *bio)
 	struct bio_vec *bv;
 	int i;
 
+	if (trace_android_fs_dataread_end_enabled() &&
+	    (bio_data_dir(bio) == READ)) {
+		struct page *first_page = bio->bi_io_vec[0].bv_page;
+
+		if (first_page != NULL)
+			trace_android_fs_dataread_end(first_page->mapping->host,
+						      page_offset(first_page),
+						      bio->bi_iter.bi_size);
+	}
+
 	bio_for_each_segment_all(bv, bio, i) {
 		struct page *page = bv->bv_page;
 		page_endio(page, bio_op(bio),
@@ -60,6 +78,24 @@ static void mpage_end_io(struct bio *bio)
 
 static struct bio *mpage_bio_submit(int op, int op_flags, struct bio *bio)
 {
+	if (trace_android_fs_dataread_start_enabled() && (op == REQ_OP_READ)) {
+		struct page *first_page = bio->bi_io_vec[0].bv_page;
+
+		if (first_page != NULL) {
+			char *path, pathbuf[MAX_TRACE_PATHBUF_LEN];
+
+			path = android_fstrace_get_pathname(pathbuf,
+						    MAX_TRACE_PATHBUF_LEN,
+						    first_page->mapping->host);
+			trace_android_fs_dataread_start(
+				first_page->mapping->host,
+				page_offset(first_page),
+				bio->bi_iter.bi_size,
+				current->pid,
+				path,
+				current->comm);
+		}
+	}
 	bio->bi_end_io = mpage_end_io;
 	bio_set_op_attrs(bio, op, op_flags);
 	guard_bio_eod(op, bio);

include/trace/events/android_fs.h

@@ -0,0 +1,65 @@
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM android_fs
+
+#if !defined(_TRACE_ANDROID_FS_H) || defined(TRACE_HEADER_MULTI_READ)
+#define _TRACE_ANDROID_FS_H
+
+#include <linux/tracepoint.h>
+#include <trace/events/android_fs_template.h>
+
+DEFINE_EVENT(android_fs_data_start_template, android_fs_dataread_start,
+	TP_PROTO(struct inode *inode, loff_t offset, int bytes,
+		 pid_t pid, char *pathname, char *command),
+	TP_ARGS(inode, offset, bytes, pid, pathname, command));
+
+DEFINE_EVENT(android_fs_data_end_template, android_fs_dataread_end,
+	TP_PROTO(struct inode *inode, loff_t offset, int bytes),
+	TP_ARGS(inode, offset, bytes));
+
+DEFINE_EVENT(android_fs_data_start_template, android_fs_datawrite_start,
+	TP_PROTO(struct inode *inode, loff_t offset, int bytes,
+		 pid_t pid, char *pathname, char *command),
+	TP_ARGS(inode, offset, bytes, pid, pathname, command));
+
+DEFINE_EVENT(android_fs_data_end_template, android_fs_datawrite_end,
+	TP_PROTO(struct inode *inode, loff_t offset, int bytes),
+	     TP_ARGS(inode, offset, bytes));
+
+#endif /* _TRACE_ANDROID_FS_H */
+
+/* This part must be outside protection */
+#include <trace/define_trace.h>
+
+#ifndef ANDROID_FSTRACE_GET_PATHNAME
+#define ANDROID_FSTRACE_GET_PATHNAME
+
+/* Sizes an on-stack array, so careful if sizing this up ! */
+#define MAX_TRACE_PATHBUF_LEN	256
+
+static inline char *
+android_fstrace_get_pathname(char *buf, int buflen, struct inode *inode)
+{
+	char *path;
+	struct dentry *d;
+
+	/*
+	 * d_obtain_alias() will either iput() if it locates an existing
+	 * dentry or transfer the reference to the new dentry created.
+	 * So get an extra reference here.
+	 */
+	ihold(inode);
+	d = d_obtain_alias(inode);
+	if (likely(!IS_ERR(d))) {
+		path = dentry_path_raw(d, buf, buflen);
+		if (unlikely(IS_ERR(path))) {
+			strcpy(buf, "ERROR");
+			path = buf;
+		}
+		dput(d);
+	} else {
+		strcpy(buf, "ERROR");
+		path = buf;
+	}
+	return path;
+}
+#endif

include/trace/events/android_fs_template.h

@@ -0,0 +1,64 @@
+#if !defined(_TRACE_ANDROID_FS_TEMPLATE_H) || defined(TRACE_HEADER_MULTI_READ)
+#define _TRACE_ANDROID_FS_TEMPLATE_H
+
+#include <linux/tracepoint.h>
+
+DECLARE_EVENT_CLASS(android_fs_data_start_template,
+	TP_PROTO(struct inode *inode, loff_t offset, int bytes,
+		 pid_t pid, char *pathname, char *command),
+	TP_ARGS(inode, offset, bytes, pid, pathname, command),
+	TP_STRUCT__entry(
+		__string(pathbuf, pathname);
+		__field(loff_t,	offset);
+		__field(int,	bytes);
+		__field(loff_t,	i_size);
+		__string(cmdline, command);
+		__field(pid_t,	pid);
+		__field(ino_t,	ino);
+	),
+	TP_fast_assign(
+		{
+			/*
+			 * Replace the spaces in filenames and cmdlines
+			 * because this screws up the tooling that parses
+			 * the traces.
+			 */
+			__assign_str(pathbuf, pathname);
+			(void)strreplace(__get_str(pathbuf), ' ', '_');
+			__entry->offset		= offset;
+			__entry->bytes		= bytes;
+			__entry->i_size		= i_size_read(inode);
+			__assign_str(cmdline, command);
+			(void)strreplace(__get_str(cmdline), ' ', '_');
+			__entry->pid		= pid;
+			__entry->ino		= inode->i_ino;
+		}
+	),
+	TP_printk("entry_name %s, offset %llu, bytes %d, cmdline %s,"
+		  " pid %d, i_size %llu, ino %lu",
+		  __get_str(pathbuf), __entry->offset, __entry->bytes,
+		  __get_str(cmdline), __entry->pid, __entry->i_size,
+		  (unsigned long) __entry->ino)
+);
+
+DECLARE_EVENT_CLASS(android_fs_data_end_template,
+	TP_PROTO(struct inode *inode, loff_t offset, int bytes),
+	TP_ARGS(inode, offset, bytes),
+	TP_STRUCT__entry(
+		__field(ino_t,	ino);
+		__field(loff_t,	offset);
+		__field(int,	bytes);
+	),
+	TP_fast_assign(
+		{
+			__entry->ino		= inode->i_ino;
+			__entry->offset		= offset;
+			__entry->bytes		= bytes;
+		}
+	),
+	TP_printk("ino %lu, offset %llu, bytes %d",
+		  (unsigned long) __entry->ino,
+		  __entry->offset, __entry->bytes)
+);
+
+#endif /* _TRACE_ANDROID_FS_TEMPLATE_H */