Red Hat Bugzilla – Bug 14553
Subtile bug causing stacktrace
Last modified: 2008-05-01 11:37:56 EDT
-low memory machine (boot with linux mem=16M) -select workstation -manual partition -yes enable swap -hmm, disks are to small anyways I want a custom install -back -select no on enable swap (see bug 14551) -note going forward now again, selecting yes again for enable swap (so swap would be enabled twice) works, and there is no message about formatting swap so I guess this stepped is skipped the second time, but then why still show the box? -go back all the way to the install type selection screen -select custom install -use diskdruid -select yes for enable swap, now really tries to enable it twice -> BOOM Also see the soon to be attached stacktrace.
Created attachment 1492 [details] stacktrace
This defect is considered MUST-FIX for Winston Release-Candidate #1
This should work if you get your space setup properly before you get to the 'not enough space screen'. Trying to enable/disable swap many times is going to break.
Why? I don't see what is the problem, granted disabling swap on back is ugly, so keep a few global vars (one per partition) if swap is on on that partition, maybe this info is even somewhere in proc. If swap is in use and someone decides to change the partition table the sync will fail, this should bring up a you need to reboot screen, I guess this is already there, otherwise I forsee big problems with hd installs.
Please note that I did not go al the way up to the not enough free diskspace message, immediatly after the swap is needed now screen I went back, since I changed my mind and wanted a custom install so the problem track is: -Standard workstation install -partition -low on memory need to enable swap now -hmm I changed my mind go back and select custom -partitiom -low on memory message -BOOM So the problem is that the swap space status is remembered seperate for an custom and a workstation install, going back and forward 100 times in just install method works fine.