Re: Patch for **MAJOR** HFS.FST Bug

[harold.h@genie.com (Harold Hislop)]

Jawaid, (Dave Lyons too, if you're reading),
 
MANY thanks to both of you (and all others involved of course :) in finding
and nailing that bug in the logical/physical HFS block calculation code!
 
It's been asked here (and on GEnie as well) if applying the patch and then
accessing an existing HFS volume could have problems. In my opinion, the
answer to this is a very emphatic YES.
 
Consider... The original routine (pre-patch) would have been responsible
for calculating a block number used during writing of a file on that
volume, but after patching that routine is not likely to produce the
same results (infact hopefully it will NOT ;) This means that there is
the probability that subsequent accesses to the file in question will
fail, because the 'wrong' (actually right :) block number will be used.
 
The only =safe= way I can conceive of to get around this is to make a
complete file-by-file (NOT image!) backup of all existing HFS volumes,
apply the patch, re-initialize (high level of course) the HFS volumes,
and then restore the files from the backups. This is, IMHO, especially
critical as there aren't any HFS repair / recovery utilities for the
IIgs at this time. (tis always better to be real safe, than real sorry)
 
FWIW, Pre-patch I was able to reliably and repeatable cause corruption
of HFS volumes between 64Mb and 100Mb by placing fewer than 50 files
on the volume and then drag moving those files, repeatedly, between
subdirectories on that volume. The more files dragged at one time seemed
to have a direct impact on how soon the volume would become corrupted,
working with 20 files at a time this would occur within 30 such
operations for me.
 
After applying the patch I have =NOT= been able to duplicate the above.
Not even =ONCE=, even when bumping the number of files up to more than
500, and dragging upwards of 100 files at a time.
 
Gentlemen, I do beleive you've nailed one nasty bug, THANK YOU BOTH!
 
-Harold Hislop
Hardware Coordinator,
Apple II RoundTable, GEnie
14/32 TWGS, 1Mb Rev-C RamFast, Prototype ROM_03 in tower

--
-Harold
Hardware Coordinator,
Apple II RoundTable, GEnie
SCSI Spoken Here