> OK, I've had a hard day, and I'm trying to wind down, but this
> DOES NOT COMPUTE :-)

Sorry about your bad day, but you're not alone.  I've hacked this for at
least 3 days now at work (and not even doing my job) ;)

>Comments about 2.4 are either irrelevant or mistaken (i.e. you
>meant '2.6').  
> For current 2.6, parrallel libata uses /dev/sd? to
>refer to IDE drives.  It's also not yet recommended for production -
>I'm not suggesting you should drop it (people do need to test new
>developments), but if you think you _have_to_ refer to IDE drives as
>/dev/sd? to get them working properly, you probably have very
>obscure, or extremely new, hardware ?

Probably (actually now) irrelevant.  I did mean 2.4, however, the HD
with the 2.4 is out of the box so I couldn't verify much.  I could have
sworn in 2.4 I had IDE working but the drive would pause now and then to
perform swapping, flushes, who knows what.  Once I put in SATA support
(in 2.4) the drive performed as it should.  

>Most people can get IDE drives working ok as /dev/hd? at the
>moment.  Please try _not_ using parrallel libata in your kernel
>config (and refer to /dev/hda in /etc/fstab) to get something that
>works.  After that, you can play with parrallel libata if you wish
>(although the dynamic selection of device names in /etc/fstab will
>be 'interesting", which is why fedora mounts by label).

I had early on.  The inclusion of parallel libata was an experiment that
had identical results to not including it...so I didn't sweat it.  

Either way, a big

######  ####### #     #   ###
#     # #     # #     #   ###
#     # #     # #     #   ###
#     # #     # #######    #
#     # #     # #     #
#     # #     # #     #   ###
######  ####### #     #   ###

goes to me.  Here is the story behind it:

In my efforts to boot originally, the device was not seen because I
didn't include the Intel PIIX option for IDE.  While find this out, I
decided to remove "extra" stuff from the kernel...this included NIC,
audio, video, etc.  Well, by NIC, I mean full networking support.

Guess what?  The kernel sends UEvents via unix domain sockets which in
the kernel are included (guess!)...in the networking subsystem.  I'll
say it again:

######  ####### #     #   ###
#     # #     # #     #   ###
#     # #     # #     #   ###
#     # #     # #######    #
#     # #     # #     #
#     # #     # #     #   ###
######  ####### #     #   ###

Once networking (including domain sockets) along with my IDE and PIIX
options were compiled in...booting was successful.

Thanks everyone for your help!

---------------
David A. Gershman
[EMAIL PROTECTED]
"It's all about the path!" --d. gershman
-- 
http://linuxfromscratch.org/mailman/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/lfs/faq.html
Unsubscribe: See the above information page

Reply via email to