> I was probably unclear. This work is not done yet. It's going to > happen in upstream master this development cycle and be part of v2.26. > > You can read about it on the upstream mailing list.
> Hopefully once sfdisk has been rewritten, it will no longer contain > the shortcoming you've mentioned will not be part of v2.26. > > This is your chance to be pro-active and influence the upstream effort > in the right direction. ;) I might if I had a link to the upstream mailing list. There is no mention of it in the package, not even in the debian/README.source. Please don't assume that every bug submitter has the same background knowledge as the package maintainer and somehow knows where the upstream mailing list can be found. However, it is usually the job of the package maintainer to direct bug reports upstream as that avoids the need for everyone involved to have a subscription to the upstream mailing list when it's only this one issue I actually need resolved. If you could forward the report and the updated patch to upstream and tag the bug appropriately, that would be helpful. BTW: In order for me to interact with util-linux upstream, it would have been handy if the homepage URL in the Debian packaging was not a 404... $ wget http://userweb.kernel.org/~kzak/util-linux/ --2014-08-10 18:17:55-- http://userweb.kernel.org/~kzak/util-linux/ Resolving userweb.kernel.org (userweb.kernel.org)... failed: Name or service not known. wget: unable to resolve host address `userweb.kernel.org' I'm attaching an updated version of the patch for the current version of sfdisk in experimental. The files have been moved around but the bug remains. -- Neil Williams ============= http://www.linux.codehelp.co.uk/
Description: Drop calls to sync inside sfdisk Systems which use sfdisk to operate on multiple mountpoints at the same time suffer significant delays due to unnecessary calls to sync inside sfdisk.c Sync flushes the kernel buffers for all filesystems, local and remote. With controller caches and other layers, sync is not a guarantee that the data has actually been written. . util-linux (2.25-8.1) unstable; urgency=medium . * Non-maintainer upload. * Drop unnecessary calls to sync inside sfdisk (Closes: #755806) Author: Neil Williams <[email protected]> Bug-Debian: http://bugs.debian.org/755806 --- --- util-linux-2.25.orig/disk-utils/sfdisk.c +++ util-linux-2.25/disk-utils/sfdisk.c @@ -750,7 +750,6 @@ reread_ioctl(int fd) { static int reread_disk_partition(char *dev, int fd) { fflush(stdout); - sync(); if (is_blkdev(fd)) { printf(_("Re-reading the partition table ...\n")); @@ -3196,7 +3195,6 @@ do_fdisk(char *dev) { "to zero the first 512 bytes: dd if=/dev/zero of=/dev/foo7 bs=512 count=1\n" "(See fdisk(8).)")); - sync(); /* superstition */ }
signature.asc
Description: PGP signature
