[PATCH][man-pages] sync.2: syncfs() now returns errors if writeback fails

A patch has been merged for v5.8 that changes how syncfs() reports
errors. Change the sync() manpage accordingly.

Signed-off-by: Jeff Layton <jlayton@kernel.org>
---
 man2/sync.2 | 24 +++++++++++++++++++++++-
 1 file changed, 23 insertions(+), 1 deletion(-)

diff --git a/man2/sync.2 b/man2/sync.2
index 7198f3311b05..27e04cff5845 100644
--- a/man2/sync.2
+++ b/man2/sync.2

@@ -86,11 +86,26 @@ to indicate the error.
 is always successful.
 .PP
 .BR syncfs ()
-can fail for at least the following reason:
+can fail for at least the following reasons:
 .TP
 .B EBADF
 .I fd
 is not a valid file descriptor.
+.TP
+.B EIO
+An error occurred during synchronization.
+This error may relate to data written to any file on the filesystem, or on
+metadata related to the filesytem itself.
+.TP
+.B ENOSPC
+Disk space was exhausted while synchronizing.
+.TP
+.BR ENOSPC ", " EDQUOT
+Data was written to a files on NFS or another filesystem which does not
+allocate space at the time of a
+.BR write (2)
+system call, and some previous write failed due to insufficient
+storage space.
 .SH VERSIONS
 .BR syncfs ()
 first appeared in Linux 2.6.39;

@@ -121,6 +136,13 @@ or
 .BR syncfs ()
 provide the same guarantees as fsync called on every file in
 the system or filesystem respectively.
+.PP
+In mainline kernel versions prior to 5.8,
+.\" commit 735e4ae5ba28c886d249ad04d3c8cc097dad6336
+.BR syncfs ()
+will only fail with EBADF when passed a bad file descriptor. In 5.8
+and later kernels, it will also report an error if one or more inodes failed
+to be written back since the last syncfs call.
 .SH BUGS
 Before version 1.3.20 Linux did not wait for I/O to complete
 before returning.

-- 
2.26.2

On Wed, Jun 10, 2020 at 06:33:47AM -0400, Jeff Layton wrote:

The sentence "In mainline kernel versions prior to 5.8, syncfs() will only fail
with EBADF when passed a bad file descriptor" is ambiguous.  It could mean that
EBADF can now mean other things too.

Maybe write: "In mainline kernel versions prior to 5.8, syncfs() will only fail
when passed a bad file descriptor (EBADF)."

- Eric

On Wed, 2020-06-10 at 08:50 -0700, Eric Biggers wrote:

On Wed, Jun 10, 2020 at 06:33:47AM -0400, Jeff Layton wrote:

quoted

A patch has been merged for v5.8 that changes how syncfs() reports
errors. Change the sync() manpage accordingly.

Signed-off-by: Jeff Layton <jlayton@kernel.org>
---
 man2/sync.2 | 24 +++++++++++++++++++++++-
 1 file changed, 23 insertions(+), 1 deletion(-)

diff --git a/man2/sync.2 b/man2/sync.2
index 7198f3311b05..27e04cff5845 100644
--- a/man2/sync.2
+++ b/man2/sync.2

@@ -86,11 +86,26 @@ to indicate the error.
 is always successful.
 .PP
 .BR syncfs ()
-can fail for at least the following reason:
+can fail for at least the following reasons:
 .TP
 .B EBADF
 .I fd
 is not a valid file descriptor.
+.TP
+.B EIO
+An error occurred during synchronization.
+This error may relate to data written to any file on the filesystem, or on
+metadata related to the filesytem itself.
+.TP
+.B ENOSPC
+Disk space was exhausted while synchronizing.
+.TP
+.BR ENOSPC ", " EDQUOT
+Data was written to a files on NFS or another filesystem which does not
+allocate space at the time of a
+.BR write (2)
+system call, and some previous write failed due to insufficient
+storage space.
 .SH VERSIONS
 .BR syncfs ()
 first appeared in Linux 2.6.39;

@@ -121,6 +136,13 @@ or
 .BR syncfs ()
 provide the same guarantees as fsync called on every file in
 the system or filesystem respectively.
+.PP
+In mainline kernel versions prior to 5.8,
+.\" commit 735e4ae5ba28c886d249ad04d3c8cc097dad6336
+.BR syncfs ()
+will only fail with EBADF when passed a bad file descriptor. In 5.8
+and later kernels, it will also report an error if one or more inodes failed
+to be written back since the last syncfs call.

The sentence "In mainline kernel versions prior to 5.8, syncfs() will only fail
with EBADF when passed a bad file descriptor" is ambiguous.  It could mean that
EBADF can now mean other things too.

Maybe write: "In mainline kernel versions prior to 5.8, syncfs() will only fail
when passed a bad file descriptor (EBADF)."

- Eric

Good point. Fixed in my tree using your verbiage. I'll send out a v2
patch once I give others a chance to comment.

Thanks!
-- 
Jeff Layton [off-list ref]

Hi Jeff,

Any progress with v2 of this patch?

Thanks,

Michael


On 6/10/20 11:19 PM, Jeff Layton wrote:

On Wed, 2020-06-10 at 08:50 -0700, Eric Biggers wrote:

quoted

On Wed, Jun 10, 2020 at 06:33:47AM -0400, Jeff Layton wrote:

quoted

A patch has been merged for v5.8 that changes how syncfs() reports
errors. Change the sync() manpage accordingly.

Signed-off-by: Jeff Layton <jlayton@kernel.org>
---
 man2/sync.2 | 24 +++++++++++++++++++++++-
 1 file changed, 23 insertions(+), 1 deletion(-)

diff --git a/man2/sync.2 b/man2/sync.2
index 7198f3311b05..27e04cff5845 100644
--- a/man2/sync.2
+++ b/man2/sync.2

@@ -86,11 +86,26 @@ to indicate the error.
 is always successful.
 .PP
 .BR syncfs ()
-can fail for at least the following reason:
+can fail for at least the following reasons:
 .TP
 .B EBADF
 .I fd
 is not a valid file descriptor.
+.TP
+.B EIO
+An error occurred during synchronization.
+This error may relate to data written to any file on the filesystem, or on
+metadata related to the filesytem itself.
+.TP
+.B ENOSPC
+Disk space was exhausted while synchronizing.
+.TP
+.BR ENOSPC ", " EDQUOT
+Data was written to a files on NFS or another filesystem which does not
+allocate space at the time of a
+.BR write (2)
+system call, and some previous write failed due to insufficient
+storage space.
 .SH VERSIONS
 .BR syncfs ()
 first appeared in Linux 2.6.39;

@@ -121,6 +136,13 @@ or
 .BR syncfs ()
 provide the same guarantees as fsync called on every file in
 the system or filesystem respectively.
+.PP
+In mainline kernel versions prior to 5.8,
+.\" commit 735e4ae5ba28c886d249ad04d3c8cc097dad6336
+.BR syncfs ()
+will only fail with EBADF when passed a bad file descriptor. In 5.8
+and later kernels, it will also report an error if one or more inodes failed
+to be written back since the last syncfs call.

The sentence "In mainline kernel versions prior to 5.8, syncfs() will only fail
with EBADF when passed a bad file descriptor" is ambiguous.  It could mean that
EBADF can now mean other things too.

Maybe write: "In mainline kernel versions prior to 5.8, syncfs() will only fail
when passed a bad file descriptor (EBADF)."

- Eric

Good point. Fixed in my tree using your verbiage. I'll send out a v2
patch once I give others a chance to comment.

Thanks!

-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/

On Thu, 2020-06-25 at 11:45 +0200, Michael Kerrisk (man-pages) wrote:

Hi Jeff,

Any progress with v2 of this patch?

Thanks,

Michael

Just sent it. Thanks for the reminder!

On 6/10/20 11:19 PM, Jeff Layton wrote:

quoted

On Wed, 2020-06-10 at 08:50 -0700, Eric Biggers wrote:

quoted

On Wed, Jun 10, 2020 at 06:33:47AM -0400, Jeff Layton wrote:

quoted

A patch has been merged for v5.8 that changes how syncfs() reports
errors. Change the sync() manpage accordingly.

Signed-off-by: Jeff Layton <jlayton@kernel.org>
---
 man2/sync.2 | 24 +++++++++++++++++++++++-
 1 file changed, 23 insertions(+), 1 deletion(-)

diff --git a/man2/sync.2 b/man2/sync.2
index 7198f3311b05..27e04cff5845 100644
--- a/man2/sync.2
+++ b/man2/sync.2

@@ -86,11 +86,26 @@ to indicate the error.
 is always successful.
 .PP
 .BR syncfs ()
-can fail for at least the following reason:
+can fail for at least the following reasons:
 .TP
 .B EBADF
 .I fd
 is not a valid file descriptor.
+.TP
+.B EIO
+An error occurred during synchronization.
+This error may relate to data written to any file on the filesystem, or on
+metadata related to the filesytem itself.
+.TP
+.B ENOSPC
+Disk space was exhausted while synchronizing.
+.TP
+.BR ENOSPC ", " EDQUOT
+Data was written to a files on NFS or another filesystem which does not
+allocate space at the time of a
+.BR write (2)
+system call, and some previous write failed due to insufficient
+storage space.
 .SH VERSIONS
 .BR syncfs ()
 first appeared in Linux 2.6.39;

@@ -121,6 +136,13 @@ or
 .BR syncfs ()
 provide the same guarantees as fsync called on every file in
 the system or filesystem respectively.
+.PP
+In mainline kernel versions prior to 5.8,
+.\" commit 735e4ae5ba28c886d249ad04d3c8cc097dad6336
+.BR syncfs ()
+will only fail with EBADF when passed a bad file descriptor. In 5.8
+and later kernels, it will also report an error if one or more inodes failed
+to be written back since the last syncfs call.

The sentence "In mainline kernel versions prior to 5.8, syncfs() will only fail
with EBADF when passed a bad file descriptor" is ambiguous.  It could mean that
EBADF can now mean other things too.

Maybe write: "In mainline kernel versions prior to 5.8, syncfs() will only fail
when passed a bad file descriptor (EBADF)."

- Eric

Good point. Fixed in my tree using your verbiage. I'll send out a v2
patch once I give others a chance to comment.

Thanks!

-- 
Jeff Layton [off-list ref]