From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040116 Description of problem: "df" in Fedora1 with the kernel-smp-2.6.4-1.300 kernel does not work for some NFS filesystems: [olchansk@jam olchansk]$ df /home/olchansk Filesystem 1K-blocks Used Available Use% Mounted on df: `/home/olchansk': Value too large for defined data type [olchansk@jam olchansk]$ grep olch /proc/mounts sam:/home1/olchansk /home/olchansk nfs rw,v3,rsize=8192,wsize=8192,hard,intr,udp,lock,addr=sam 0 0 On the NFS server: [olchansk@sam olchansk]$ df /home/olchansk Filesystem 1K-blocks Used Available Use% Mounted on /home1/olchansk 30715336 27196188 3519148 89% /home/olchansk [olchansk@sam olchansk]$ grep home1 /proc/mounts /dev/hda2 /home1 reiserfs rw 0 0 This is probably a mismatch between the 2.6 kernel and "df" or "glibc", but I do not see it reported anywhere. It could be specific to NFS mouting reiserfs filesystems, too. K.O. Version-Release number of selected component (if applicable): kernel-smp-2.6.4-1.300 How reproducible: Always Steps to Reproduce: 1. blah... 2. 3. Additional info:
What does 'strace df /home/olchansk' say? It's not clear to me where this error originates.
Created attachment 99571 [details] strace df /home/olchansk
The error is coming from "statfs("/home/olchansk", 0xfeeb4fd4) = -1 EOVERFLOW (Value too large for defined data type)". The full strace is attached. K.O.
Not sure that df can really rely on any of the values then. Looks to me like reporting an error is all it can do. I wonder why statfs() returns that in the first place.
So if the problem is in "statfs", reassign the bug to the statfs bozos, but please don't just close the bug and hope the problem or I will go away. K.O.
there's not really anything that can be done to fix this that I can see. your glibc is compiled against 2.4 kernel headers. That struct changed in 2.6, so unsurprisingly, glibc gets confused and returns nonsense, which the app then tries to translate. the only answer I see is to use a glibc compiled against 2.6 headers. remember also, that FC1 never shipped with a 2.6 kernel, and as such lots of userspace isn't prepared for it. It's completely unsurprisingly that some things break.