Hi,
We received it but some of the SGI people are on holiday.
Figures :)
You didn't get errors in your log? Was you're filesystem shutdown when
this occured?
No errors or anything. Ummm. I find it difficult to read a file from a
shutdown filesystem so no, it was mounted :)
Today I went into a directory to run a program and the binary was ..
broken. Couldn't load.
You mean that the ondisk binary had changed and did not work anymore?
Bingo, since the problem persisted after I switched back (ie file still
broken).
Tried recompiling from source and that failed too, so I looked at the
source file and it was missing the end of the file.
It was my source so I gather it was the last bytes that were missing.
------
return(num_cpus);
}
------
became
------
return(num_cpu
------
People have had problems with files containing NULL characters after a
unclean shutdown but You are the first I see with a actual damaged file.
Well, null-bytes-as-data is easily explained. This I haven't seen before.
And, I checked the file and it didn't contain a few garbage characters,
the length was simply shorter. Ie my "s);\n\n" wasn't "\0\0\0" or something.
What's also odd is that I haven't gotten anything else broken in the
trip. I could expect to have broken files all over the place but if I do
I haven't found any yet.
Does the problem go away when compiling the tree with egcs? Could you
try 2.4.16 when that is merged into CVS. It might be that 2.4.15 and
-pre was not that good. It was a pre kernel after all.
I rebooted the system quite fast after I noticed the problems and I
didn't try recompiling with egcs, no. Perhaps I should, but it's just
that it's a live system. "Well, Stefan, why are you using gcc3 and a cvs
kernel on a live system?" "Duh, I donno, I'm stupid and I like bleeding
edge". :) It's my home fileserver.
Does this happen often?
Once so far. With 2.4.15-pre7. I'll be trying 2.4.16 shortly.
// Stefan
|