Floppy.x86

Now that I have a floppy drive that doesn’t appear to freeze the computer whenever I mount a disk, I thought I’d try booting Haiku for the first time since 2002.

Whenever I dd the floppy to disk, it always says 80+1 in, 80+0 out. Is that right? I do a surface scan in DriveSetup and whatever floppy I use it says is fine, but when I boot to it, I get, “Loading,” then two and a bit lines of “.”, and then it says, “Error reading disk,” or something to that effect. If I press enter it tries again and stops in the exact same place every time, whatever the disk.

What am I doing wrong?

Kev

it may be a bad floppy, have you for sure isolated that it isn’t bad media?

Wouldn’t the DriveSetup surface scan fail if the floppy was bad? Anyway, I tried two like this that don’t fail the scan, and they both fail to boot in exactly the same spot. I’ve got two other floppies I can try, though…we’ll see.

Well, my other two were bad, and didn’t pass the surface scan.

As for the good/not-so-good other two I have, I noticed this when dd’ing:

$ dd if=floppy.x86 of=/dev/disk/floppy/raw bs=18k
/bin/dd: /dev/disk/floppy/raw: Invalid argument
80+1 records in
80+0 records out

It doesn’t complain about the argument until it’s done writing, though.

Also, on boot, the number of dots is identical. I get:

.Loading…(…)
(…)
…<–stops below the ‘g’ every time on either disk

I’ll grab some more floppies next time I go home…

Kev

If you happen to have a CPU above 2.1GHz and didn’t install the cpu_fix, it may be that the floppy cannot be written correctly because of timing issues. When I first booted on my new hardware (P4 3.0GHz) I downloaded a BIOS update and wrote it to a floppy. Tracker did not report any error, but the ROM was not written completely and screwed up the new MB (but luckily there is that neat BIOS recovery function on AMI BIOSes that doesn’t require the BIOS to work at all).

Edit:
Normally you’ll get two equal numbers from dd if writing was successful. The first number is the number of whole blocks, the +1 means a partial block. This is normal when the file is not aligned to your blocksize (i.e. the floppy image’s size is not a multiple of the 18k blocksize). The upper two values mean “read”, the lower mean “written”.
If writing is really successfull, you should get that +1 in the lower line too and you should get <floppy-image-size> / 18k whole blocks.

Just upgraded to a PIII 933 MHz.

Hmm…I wonder why it’s not writing correctly. 80 x 18k = 1,440k, but floppies hold 1,474k, don’t they?

Kev

Using a boot floppy should work, but it’s not the way to test haiku now.

Really you should run the script to make a disk image and write that to a spare hard disk partition or to an image file that you can load up in bochs.

Speaking of that:

I am having a REALLY hard time getting started with “Haiku” I’ve been using BeOS for years, and I remember being near tears when I found out about the sale of IP to Palm. I want to use as much as I can of Haiku (especially the networking because I can’t get my network card working) but now the build factory is down, and I can’t find the necessary pieces for subversion, and I haven’t been able to find an image, iso, or files I need to make any of these pieces…

help please!
:slight_smile:

The original idea was to replace BeOS piece-by-piece, but it’s not necessarily possible. In their current state, most of the pieces aren’t finished enough to replace the BeOS ones. If there’s a driver you need though, they are usually made R5 compatible too.

Networking isn’t ready to be used as a replacement yet, and certainly won’t bring support for your card unless there’s a driver already (in which case it should work with the normal BeOS net server).

Well this has been a problem for me:

I have a card that is a Broadcom 4401, and I’ve downloaded the driver that was put on BeBits a while ago, and it totally doesn’t work, as far as I can tell there is no entry in devfs, and there is no indication of a failure that I can identify, and certainly there is no card in networking prefrences. I’ve tryed to install bone7a, but that seriously kills BeOS Development Edition ( there is a kernel debugging land complaint about input-server). It’s all very sad, I’m trying to use BeOS with my laptop (because I hate sitting at a desk) but I am helpless without network connectivity!

Sad, sad state of affairs!

ANY help is GREATLY apreciated.

hpux wrote:
Well this has been a problem for me:

I have a card that is a Broadcom 4401, and I’ve downloaded the driver that was put on BeBits a while ago, and it totally doesn’t work, as far as I can tell there is no entry in devfs, and there is no indication of a failure that I can identify, and certainly there is no card in networking prefrences. I’ve tryed to install bone7a, but that seriously kills BeOS Development Edition ( there is a kernel debugging land complaint about input-server). It’s all very sad, I’m trying to use BeOS with my laptop (because I hate sitting at a desk) but I am helpless without network connectivity!

Sad, sad state of affairs!

ANY help is GREATLY apreciated.

the driver was fixed for net_server support recently

BeOS hasn’t got a ‘devfs’. Its not Linux.

Quote:
the driver was fixed for net_server support recently

This is the driver that I have been using :frowning:

Quote:
BeOS hasn't got a 'devfs'. Its not Linux.

Correct, BeOS it not linux. But you should read the BeBook:
http://www.tycomsystems.com/beos/BeBook/Drivers/Intro.html

hpux wrote:
Quote:
the driver was fixed for net_server support recently

This is the driver that I have been using :frowning:

Quote:
BeOS hasn't got a 'devfs'. Its not Linux.

Correct, BeOS it not linux. But you should read the BeBook:
http://www.tycomsystems.com/beos/BeBook/Drivers/Intro.html

I’ve read it. I’ve got the physical paper versions. I’ve been using BeOS and developing on it for a long time now…

It has got a /dev structure, but its not got a Linux like devfs. However, a lot of people on here seem to assume its a Linux clone, e.g the question about kernel modules recently…

Quote:
It *has* got a /dev structure, but its not got a Linux like devfs. However, a lot of people on here seem to assume its a Linux clone, e.g the question about kernel modules recently..

Here is the section of the intro to drivers from the BeBook:

Quote:
devfs
The kernel manages device drivers through devfs, the device file system that's mounted at /dev during the boot process. In order to be accessed, a driver must "publish" itself by adding an entry in the /dev hierarchy. The basic Posix I/O functions (open(), read(), write(), readv(), writev(), ioctl(), and close()) can then be used.

Devfs makes the drivers available as needed in /dev; this usually happens the first time a program iterates through the directory entries for a subdirectory in /dev. The kernel knows where in the /dev hierarchy to publish drivers based on their location in /boot/beos/system/add-ons/kernel/drivers/dev. For example, the ATAPI driver publishes drivers in /dev/disk/ide/atapi, the driver is located in /boot/beos/system/add-ons/kernel/drivers/dev/disk/ide/atapi. Whew.

You can see this device hierarchy by using the "ls" command from a Terminal window. "ls /dev" will show you the root of the device hierarchy, "ls /dev/disk" will show you disk device busses, "ls /dev/disk/ide" will show you the IDE devices, and so forth.

In reality, drivers tend to publish themselves in multiple locations in the /dev hierarchy, so instead of putting duplicate copies of the driver in the .../drivers/dev tree, the driver binaries are put at /boot/beos/system/add-ons/kernel/drivers/bin, and symlinks are created in the .../drivers/dev tree at the appropriate place. (The same is also done for drivers in /boot/home/config/add-ons/kernel/drivers/....) </blockquote></div>

It does have a dev structure, and indeed a devfs. This is the system with which the driver hooks are accessed. (this is from the “writing drivers” section of the BeBook) It would seem to follow, then, that the devfs is a relatively important link in the driver chain. (so to speak)

I do want to emphsize that I am in NO way under the impression that BeOS has anything to do with linux, and that is what I love about it. It is, in fact superior to linux in many ways, especially in the context of a Media OS / Desktop OS. I cannot speak to what others have said about linux kernel modules, as they are not my comments. I just want to know why the driver for my network doesn’t work! :slight_smile: Although the origional code for this driver was a linux driver with BeOS hooks. This, I think is ingenious. BeOS developers can leverage the immense library of open source drivers to vastly expand their hardware support. From what I so far understand BeOS provides a nice, and developer-friendly interface for drivers and add-ons.