Re: zeroes read back more often than appended

[Pat LaVarre <p.lavarre@ieee.org>]

I first saw udf.ko of loop.ko read more zeroes than appended in RedHat
2.6.0-test6 on one pc described by dmesg as:
495MB LOWMEM available.
CPU0: Intel(R) Pentium(R) 4 CPU 2.40GHz stepping 09
CPU1: Intel(R) Pentium(R) 4 CPU 2.40GHz stepping 09
..... host bus clock speed is 199.0492 MHz.

I intend to try a number of 2.6 pc's, to see if I get consistent
results.

But initially just now, I thought I should mention, I see correct '\xAA'
read back, no troublesome zeroes, in Knoppix 2.4.22-xfs on a third pc,
in 30..90 s/test, with a loop on ext2, on one pc described by dmesg as:
255MB LOWMEM available.
Detected 866.394 MHz processor.
CPU0: Intel Pentium III (Coppermine) stepping 03
..... host bus clock speed is 133.2875 MHz.

As I repeated this test, I regretfully noticed that my original
instructions did not explicitly include such further hints as:

sudo mkdir /mnt/loop0

# visit http://sourceforge.net/projects/linux-udf/
cp -ip udftools-1.0.0b2/mkudffs/mkudffs ~/bin

Also machines with less calloc'able virtual memory and/or low quotas
can't demo that a single fwrite reads more zeroes than appended:

udfwh mkudffs 0x7800000 0x1
udfwh mkudffs 0xFFFFFFF 0x1

Instead because of those limits on such machines we can only repeat the
themultiple fwrite tests e.g.

udfwh mkudffs 0xFFFFFF 0xB
udfwh mkudffs 0xFFFF 0xC00

Pat LaVarre