Christoph Hellwig <hch@xxxxxx> writes:
> fsync is required for data integrity as there is no gurantee that
> data makes it to disk at any specified time without it. Even for
> ext3 with data=ordered mode the file system will only commit all
> data at some point in time that is not guaranteed.
It comes from this one:
commit aafe9fbaf4f1d1f27a6f6e3eb3e246fff81240ef
Author: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
Date: Wed Jun 18 15:18:44 2008 -0700
Add config option to enable 'fsync()' of object files
As explained in the documentation[*] this is totally useless on
filesystems that do ordered/journalled data writes, but it can be a
useful safety feature on filesystems like HFS+ that only journal the
metadata, not the actual file contents.
It defaults to off, although we could presumably in theory some day
auto-enable it on a per-filesystem basis.
[*] Yes, I updated the docs for the thing. Hell really _has_ frozen
over, and the four horsemen are probably just beyond the horizon.
EVERYBODY PANIC!
> diff --git a/Documentation/config.txt b/Documentation/config.txt
> index 0e25b2c92..9a1cec5c8 100644
> --- a/Documentation/config.txt
> +++ b/Documentation/config.txt
> @@ -866,10 +866,8 @@ core.whitespace::
> core.fsyncObjectFiles::
> This boolean will enable 'fsync()' when writing object files.
> +
> -This is a total waste of time and effort on a filesystem that orders
> -data writes properly, but can be useful for filesystems that do not use
> -journalling (traditional UNIX filesystems) or that only journal metadata
> -and not file contents (OS X's HFS+, or Linux ext3 with "data=writeback").
> +This option is enabled by default and ensures actual data integrity
> +by calling fsync after writing object files.
I am somewhat sympathetic to the desire to flip the default to
"safe" and allow those who know they are already safe to tweak the
knob for performance, and it also makes sense to document that the
default is "true" here. But I do not see the point of removing the
four lines from this paragraph; the sole effect of the removal is to
rob information from readers that they can use to decide if they
want to disable the configuration, no?
