Related link: http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=115980

UPDATE: A commenter kindly pointed me to the equivalent Mandrake 10 bug. From what I can read it seems the handling of this problem in Mandrake has been as worrisome as that in FC2. The bug is marked as “fixed” although comments make it clear that the mooted workaround is not sufficient for some users. As such I extend the same warning to those considering any version of Mandrake 10.0. Be very careful if you have a dual XP boot set-up.

UPDATE: See this article quoting a response from a Fedora developer.

I have used Red Hat since 4.2, and it has been my distro of choice. I’ve earnestly tried Debian, Mandrake, Gentoo and SuSE at various times, but I’ve always come back to Red Hat, probably not because of any of the silly reasons you see bandied about in flame wars, but rather because I’m already comfortable and familar with Red Hat. When the natural evolution of Red Hat took me to Fedora Core 1, I followed and have been very happy so far.

I’ve been eyeing the progress of FC2, which includes Kernel 2.6 and other goodies. I tried the third and final FC2 test release but had problems with the installer that were already listed in bugzilla and decided to try again upon the final release, scheduled for this week. Before doing so I browsed the mailing list to see what success others were having and came across this posting, from which I quote:


A very serious bug exists which can render dual-boot Windows XP
installations inoperable. The problem has been discussed in other
threads here on this list, and can be found in bugzilla, so I won’t go
over it in more detail. Please see bugzilla bugs 120128 and 115980.
This problem happens on my hardware even if the harddrive is wiped, and
a fresh Windows XP install is performed before FC2-test3.

Note: bug #120128 is a dupe of #115980.

This bug is still acknowledged as unfixed on the eve of the final FC2 release. On reading further, several things alarmed me:

  1. The attitude of many in the thread who violently flamed those who were raising the alarm
  2. The overall silence of core developers and release notes on a matter of such magnitude
  3. The very fact that such a major distribution does not treat a widely reproduced bug with serious consequences of data loss as a show-stopper, or at least a code-red priority

Before commenting further I do want to make sure I reiterate that people see the actual bugzilla item for the raw facts as they’re known. Also see this thread which is a bit more focused on solutions and narrowing down the problem than the flame war over whether the Fedora community is taking this bug too lightly.

There has been some discussion as to whether this is actually a bug in Grub, in Kernel 2.6 (some say they came across is in Mandrake 10), or in Windows XP. None of the uncertainty over root cause excuses a responsible distribution from treating any such issue with the utmost seriousness. This bug could cause a significant cross-section of users to lose data, and so at the very least it should be affixed with a clarion horn, bold letters and blink tags in the various release notes and announcements. Instead, the sorts of discussion I see about this situation include jibes at dual-booters and even snide suggestions that FC2 is doing users a favor by neutralizing their Windows paritions. I can be as spiky in a flame war as anyone else, but I’m very surprised at how a matter that could so seriously damage the reputation of Fedora is being treated with so little apparent seriousness by all but those who have themselves lost time and data.

Speaking for myself, I have two computers on which I’d planned to make the upgrade. One (my laptop) is a dual-boot to XP. Yes, I usually back up before such upgrades and installations, but I don’t have the time to go through the reinstall-XP-then-restore dance so I’m not taking the chance on my laptop and will leave it at FC1 until there is a fix for this issue. I shall upgrade the Linux-only box right away and, I hope, not run into an other unpleasant surprises.

The most important matter is fixing the bug. Do you have any new data that might be useful in doing so?