Re: [owl-users] SATA controller

From: vlad (faktroot@private)
Date: Tue Nov 21 2006 - 20:59:19 PST

Hello Solar Designer!

Wednesday, November 22, 2006, 3:55:59 AM, you wrote:

>> Unfortunately, your advice have not brought due result. :(
SD> This is because you've failed to follow some of the instructions that
SD> were given to you, please see below:

Probably you are right also I something has overlooked, though
tried to make everything, that to me recommended

>> At attempt to load for the controller I receive the module
>> Numerous abuse on "unresolved symlink"
SD> You must have meant "unresolved symbol", not "symlink".  This suggests
SD> that you did not apply some of the patches from under /usr/src/kernel as
SD> I had suggested in another posting.  In order for you to be able to load
SD> modules into the CD-booted kernel, those modules have to be from a
SD> kernel patched and configured in _exactly_ the same way as the kernel on
SD> the CD.  This means that you must apply all of the patches under
SD> /usr/src/kernel (for Owl 2.0, that's linux-2.4.32-ow1 and
SD> patch-cryptoloop-jari-2.4.32-ow1) _and_ use /boot/.config off the CD
SD> (your enabling the compilation of some drivers as modules must be the
SD> only change).

Yes, you are absolutely right, I have mixed the message on a mistake.
But I really at first have established all patches on source texts of
a kernel. And modules for the controller of a parent payment are
available only for versions of a kernel 2.4.18 and 2.4.20.
And when I try to compile and load them I receive the message,
that it not that version of a kernel. :(
The order of my actions such:
I have remounted /tmp and have increased it up to 400 Mb
# mount tmpfs /tmp -t tmpfs -osize=400M
And also has remounted /usr/src
Also has create the symlynk /tmp/kernel/linux-2.4.32/usr/src/linux
Has applied patches linux-2.4.32-ow1 and patch-cryptoloop-jari-2.4.32-ow1
Has copied /boot/.config in a tree of source codes
# make menuconfig
Has included emulation SCSI in the form of the module, support SATA, controllers
# make dep; make clean; make bzImage; make modules
I load the module of emulation SCSI.
And then, when I try to load modules of the controller I receive

>> The question (partially) was solved switching of a disk on other controller of a same payment,
>> Which has earned at once, but has given speed of work with a disk
>> hdparm-t/dev/hdb1 ~ 3.3 MB/sec
>> And it on a SATA-disk!
SD> Well, the name /dev/hdb1 suggests that your system sees this as an IDE
SD> drive, and this low transfer rate is normal for 16-bit PIO mode (this
SD> happens when you have an IDE controller - or one pretending to be IDE -
SD> for which there's no driver in the running kernel).
>> But the most interesting, it after installation of system on a disk,
>> I at once could load the necessary module for the first controller!
SD> I am guessing that you had used your newly-built kernel for this
SD> installed system.  So you finally had the kernel and the module match.

You are absolutely right, I had to specify in BIOS what to consider
SATA-devices on the given controller as IDE-disks.
Then, when I have made a monolithic kernel with support of my devices
and was loaded from it speed of work with a disk became 56-58 Mb/sec

If I do any inexcusable mistakes I ask to not judge me strictly.
Before I used and I use on servers of Linux-system which use
other approach in installation.
It Trustix Secure Linux, RedHat, Slackware.
But always wished to try in work Openwall Linux.
Now I study system. It all more and is pleasant to me more. :)
Certainly there are many questions which I try to solve all over again,
And if it is impossible - to address to community.

Big all thanks for the help!

With best wishes,
 vlad                            mailto:faktroot@private

System/Network Administrator
TV/Video Engineer
ICQ 163227020

Все животные, кроме человека, знают, что главное дело жизни - наслаждаться ею.


This archive was generated by hypermail 2.1.3 : Tue Nov 21 2006 - 21:00:56 PST