I think this problem is within the memory management
system and/or kswapd.
I have 0.5GB RAM along with about the same amount of
swap, and when I wrote a program that just malloced
itself more and more memory, it did "Oops 0" within
kswapd. The address it gave for the PC is nonsense.
The malloc was done in (1024*1024-80) byte sets, plus
each kilobyte (every 1024 bytes) address was stored a byte
in. The way the Alpha codes this is to have a fetch of
an integer, then some bit/byte-stuffing, and storing that
integer back.
In the first run the blowup happened in fairly low
memory usage situation, even before it got to swap
at all! In the second occasion I ran the program
right after boot, and it went well into swap, and
oopsed when no more swap was free.
In the first case the virtual address referred at
some completely garbled address, the next time it
was all zeroes.
All of this with 2.1.125pre2 (RAID-code patch against
2.1.124 can not be applied cleanly to any latter one..)
> Will check out 2.1.125ac3 later hopefully.
> Chris
/Matti Aarnio <matti.aarnio@sonera.fi>
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/