> Nate Eldredge <nate@cartsys.com> writes:
>
> > I'm chasing a bug which I found while doing a stress-test with `(tar x; tar
> > d) &' times seventeen. It involves occasional corruption of an extracted
> > file (every few days). [...] Is there anything in tar or the kernel
> > that handles 32-byte chunks and might be able to cause such corruption?
> > Alternatively, has anyone heard of similar problems with the K6-3?
>
> I'm not at all claiming that `tar' is innocent, but from your description,
> I do not remember another user report, or a known bug, that would explain
> the above behaviour. Of course, if you get some information that allows
> to strongly suspect `tar', we should surely investigate. Keep happy!
I do not think is is `tar' but the CPU-chipset combo plus egcs compiled
for Intel and not for AMD. I have not had time to reproduce the test,
but the FIC 503+ MVP3, BIOS 115JE35s with AMD K6-2/450 3D-Now on a stock
Slackware 3.5 install would eat/barf/grow the root partition.
Now that very same board revision (same shipping lot) with an Intel 233
MMX work correctly. One should note that egcs+MVP3+K6-2/450 barfed and
puked "gcc error 11" 54 times in trying to compile a 2.2.X kernel.
HOWEVER, drop in the old Intel jobber and flawless compiles.
Bus 0, device 0, function 0:
Host bridge: VIA Technologies VT 82C597 Apollo VP3 (rev 4).
Medium devsel. Fast back-to-back capable. Master Capable. Latency=16.
Prefetchable 32 bit memory at 0xe0000000 [0xe0000008].
Bus 0, device 1, function 0:
PCI bridge: VIA Technologies VT 82C598 Apollo MVP3 AGP (rev 0).
Medium devsel. Master Capable. No bursts. Min Gnt=4.
Bus 0, device 7, function 0:
ISA bridge: VIA Technologies VT 82C586 Apollo ISA (rev 71).
Medium devsel. Master Capable. No bursts.
Bus 0, device 7, function 1:
IDE interface: VIA Technologies VT 82C586 Apollo IDE (rev 6).
Medium devsel. Fast back-to-back capable. Master Capable. Latency=64.
I/O at 0xe400 [0xe401].
Note that 82C586 revision exceeds the 0x20 or rev 32, UDMA minimum version
limit for support, regardless of host bridge. I just found this trivia the
other night and decideing what to do with it..........
Cheers............
Andre Hedrick
The Linux IDE guy
-
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/