GNU bug report logs - #7421
[Feature request]: add option to dd to issue ioctl(BLKFLSBUF) on output descriptior after each write or at the end

Previous Next

To add a comment to this bug, you must first unarchive it, by sending
a message to control AT debbugs.gnu.org, with unarchive 7421 in the body.
You can then email your comments to 7421 AT debbugs.gnu.org in the normal way.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Марк Коренберг <socketpair <at> gmail.com>
To: bug-coreutils <at> gnu.org
Subject: [Feature request]: add option to dd to issue ioctl(BLKFLSBUF) on
	output descriptior after each write or at the end
Date: Wed, 17 Nov 2010 15:41:17 +0500

[Feature request]: add option to dd to issue ioctl(BLKFLSBUF) on
output descriptior after each write or at the end

I already has sent a message about fsync/fdatasync after each write.
It seems that ioctl(BLKFLSBUF) need be implemented in same semantics.

in oflags and conv
i.e. ioctl after each write oflags=blkflsbuf
and ioctl at the end, if specified in conv=blkflsbuf

-- 
Segmentation fault

Message #8 received at 7421 <at> debbugs.gnu.org (full text, mbox):

From: Pádraig Brady <P <at> draigBrady.com>
To: Марк Коренберг
 <socketpair <at> gmail.com>, 7421 <at> debbugs.gnu.org
Subject: Re: bug#7421: [Feature request]: add option to dd to issue
 ioctl(BLKFLSBUF) on output descriptior after each write or at the end
Date: Sun, 25 Jan 2015 18:41:52 +0000

unarchive 7421
tag 7421 wontfix
close 7421
stop

On 17/11/10 10:41, Марк Коренберг wrote:
> [Feature request]: add option to dd to issue ioctl(BLKFLSBUF) on
> output descriptior after each write or at the end
> 
> I already has sent a message about fsync/fdatasync after each write.
> It seems that ioctl(BLKFLSBUF) need be implemented in same semantics.
> 
> in oflags and conv
> i.e. ioctl after each write oflags=blkflsbuf
> and ioctl at the end, if specified in conv=blkflsbuf

This is already supported with the `blockdev --flushbufs` command.
Given this is a low level linux specific interface, it's more suited
to the blockdev command than a more general tool like dd.
Note ioctl(BLKFLSBUF) only writes out dirty pages to the block device,
it doesn't guarantee to send a flush request to the device.
Also http://lwn.net/Articles/433384/ mentions that BLKFLSBUF also
invalidates the bdev mapping, which isn't generally desirable,
and doesn't work for non-block file systems.

We're considering adding syncfs() support to the sync command
which should cater for much of the use case you describe here:
That's discussed at http://bugs.gnu.org/19681

thanks,
Pádraig.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.