History

fs/sync.c: sync_file_range(2) may use WB_SYNC_ALL writeback

23d0127096cb ("fs/sync.c: make sync_file_range(2) use WB_SYNC_NONE writeback") claims that sync_file_range(2) syscall was "created for userspace to be able to issue background writeout and so waiting for in-flight IO is undesirable there" and changes the writeback (back) to WB_SYNC_NONE. This claim is only partially true. It is true for users that use the flag SYNC_FILE_RANGE_WRITE by itself, as does PostgreSQL, the user that was the reason for changing to WB_SYNC_NONE writeback. However, that claim is not true for users that use that flag combination SYNC_FILE_RANGE_{WAIT_BEFORE|WRITE|_WAIT_AFTER}. Those users explicitly requested to wait for in-flight IO as well as to writeback of dirty pages. Re-brand that flag combination as SYNC_FILE_RANGE_WRITE_AND_WAIT and use WB_SYNC_ALL writeback to perform the full range sync request.

Amir Goldstein committed 7 years ago

Commit c553ea41 parent 53269052 files · +18 −6Browse files

Unified · Split

2 files
  1. fs/sync.c +15 −6
  2. include/uapi/linux/fs.h +3 −0
fs/sync.c+15−6
@@ -292,8 +292,14 @@ int sync_file_range(struct file *file, loff_t offset, loff_t nbytes, · 291 unmodified lines
292 }292 }
293 293
294 if (flags & SYNC_FILE_RANGE_WRITE) {294 if (flags & SYNC_FILE_RANGE_WRITE) {
295 int sync_mode = WB_SYNC_NONE;
296
297 if ((flags & SYNC_FILE_RANGE_WRITE_AND_WAIT) ==
298 SYNC_FILE_RANGE_WRITE_AND_WAIT)
299 sync_mode = WB_SYNC_ALL;
300
295 ret = __filemap_fdatawrite_range(mapping, offset, endbyte,301 ret = __filemap_fdatawrite_range(mapping, offset, endbyte,
296 WB_SYNC_NONE);302 sync_mode);
297 if (ret < 0)303 if (ret < 0)
298 goto out;304 goto out;
299 }305 }
@@ -306,9 +312,9 @@ out: · 6 unmodified lines
306}312}
307 313
308/*314/*
309 * sys_sync_file_range() permits finely controlled syncing over a segment of315 * ksys_sync_file_range() permits finely controlled syncing over a segment of
310 * a file in the range offset .. (offset+nbytes-1) inclusive. If nbytes is316 * a file in the range offset .. (offset+nbytes-1) inclusive. If nbytes is
311 * zero then sys_sync_file_range() will operate from offset out to EOF.317 * zero then ksys_sync_file_range() will operate from offset out to EOF.
312 *318 *
313 * The flag bits are:319 * The flag bits are:
314 *320 *
@@ -325,7 +331,7 @@ out: · 10 unmodified lines
325 * Useful combinations of the flag bits are:331 * Useful combinations of the flag bits are:
326 *332 *
327 * SYNC_FILE_RANGE_WAIT_BEFORE|SYNC_FILE_RANGE_WRITE: ensures that all pages333 * SYNC_FILE_RANGE_WAIT_BEFORE|SYNC_FILE_RANGE_WRITE: ensures that all pages
328 * in the range which were dirty on entry to sys_sync_file_range() are placed334 * in the range which were dirty on entry to ksys_sync_file_range() are placed
329 * under writeout. This is a start-write-for-data-integrity operation.335 * under writeout. This is a start-write-for-data-integrity operation.
330 *336 *
331 * SYNC_FILE_RANGE_WRITE: start writeout of all dirty pages in the range which337 * SYNC_FILE_RANGE_WRITE: start writeout of all dirty pages in the range which
@@ -337,10 +343,13 @@ out: · 5 unmodified lines
337 * earlier SYNC_FILE_RANGE_WAIT_BEFORE|SYNC_FILE_RANGE_WRITE operation to wait343 * earlier SYNC_FILE_RANGE_WAIT_BEFORE|SYNC_FILE_RANGE_WRITE operation to wait
338 * for that operation to complete and to return the result.344 * for that operation to complete and to return the result.
339 *345 *
340 * SYNC_FILE_RANGE_WAIT_BEFORE|SYNC_FILE_RANGE_WRITE|SYNC_FILE_RANGE_WAIT_AFTER:346 * SYNC_FILE_RANGE_WAIT_BEFORE|SYNC_FILE_RANGE_WRITE|SYNC_FILE_RANGE_WAIT_AFTER
347 * (a.k.a. SYNC_FILE_RANGE_WRITE_AND_WAIT):
341 * a traditional sync() operation. This is a write-for-data-integrity operation348 * a traditional sync() operation. This is a write-for-data-integrity operation
342 * which will ensure that all pages in the range which were dirty on entry to349 * which will ensure that all pages in the range which were dirty on entry to
343 * sys_sync_file_range() are committed to disk.350 * ksys_sync_file_range() are written to disk. It should be noted that disk
351 * caches are not flushed by this call, so there are no guarantees here that the
352 * data will be available on disk after a crash.
344 *353 *
345 *354 *
346 * SYNC_FILE_RANGE_WAIT_BEFORE and SYNC_FILE_RANGE_WAIT_AFTER will detect any355 * SYNC_FILE_RANGE_WAIT_BEFORE and SYNC_FILE_RANGE_WAIT_AFTER will detect any
include/uapi/linux/fs.h+3−0
@@ -320,6 +320,9 @@ struct fscrypt_key { · 319 unmodified lines
320#define SYNC_FILE_RANGE_WAIT_BEFORE 1320#define SYNC_FILE_RANGE_WAIT_BEFORE 1
321#define SYNC_FILE_RANGE_WRITE 2321#define SYNC_FILE_RANGE_WRITE 2
322#define SYNC_FILE_RANGE_WAIT_AFTER 4322#define SYNC_FILE_RANGE_WAIT_AFTER 4
323#define SYNC_FILE_RANGE_WRITE_AND_WAIT (SYNC_FILE_RANGE_WRITE | \
324 SYNC_FILE_RANGE_WAIT_BEFORE | \
325 SYNC_FILE_RANGE_WAIT_AFTER)
323 326
324/*327/*
325 * Flags for preadv2/pwritev2:328 * Flags for preadv2/pwritev2:

Keyboard shortcuts

General

Show keyboard shortcuts
?
Focus the current search
/

Issues and pull requests

Open filters
f

Issues, pull requests and Inbox

Select next item
j↓
Select previous item
k↑
Open selected item
Enter
Dismiss filters or clear selection
Esc

Repository code

Go to file
t