Media Check errors on Fedora Core 6

[Mike A. Harris]

Mike A. Harris wrote:

Hate to reply to myself, but I accidentally left off a few details...

> As a hopefully useful additional datapoint...
> 
> I've done very similar, but with the DVD image, having done:
> 
> 1) Downloaded DVD ISO of FC6 via bittorrent
> 
> 2) Verified the SHA1SUM of the image
> 
> 3) Burned it to DVD+R and used the verification feature of the burn
>    software to ensure the image burned correctly.
 >
> 4) Booted off DVD and attempted to install over top of an existing FC5
>    install.

After booting off DVD, I did a mediacheck which passed successfully.

> The first attempt got as far as bringing up disk druid where I planned
> to custom partition.  The installation hung with a kernel BUG()
> assertion which was dumped to one of the consoles.  Anaconda et al. was
> no longer functional, however VC switching still worked.
> 
> The second attempt got slightly further, allowing me to select my
> partitioning scheme in disk druid, however when I selected "Next" to
> leave disk druid and proceed, the installation hung again, this time
> with a fairly large python traceback.

There were also indications from the kernel that it was getting disk
errors from the DVD drive.  Booting into FC5, and creating a new ISO
by ripping from the burned DVD, I was able to compare the original
downloaded ISO to the one I recopied from my burned DVD, and they were
identical.  This confirms that my burn was unflawed.

> Unfortunately, I didn't have my digital camera handy at the time, and
> wasn't able to capture the details for a proper bug report yet.  When
> time permits, I'm going to attempt a 3rd install with camera in hand
> and try to capture useful details to dump in bugzilla.  For the mean
> time, I thought it might be useful to share this here in case it helps
> others pinpoint down any problems.
> 
> This system is a 2.8Ghz Intel Pentium 4 Northwood, with an i865 chipset,
> PATA drives.  Currently running FC5 and another OS without trouble.
> 
> I'll dump more info into bugzilla when I can reproduce it again.

I suspect the problem is kernel IDE driver related bugs, or perhaps
related to DMA.  Either way, all signs point to kernel bugs and not
hardware bugs nor bad burn.