[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Interface for SCSI transactions ?

From: olafBuddenhagen
Subject: Re: Interface for SCSI transactions ?
Date: Wed, 31 Aug 2011 06:01:20 +0200
User-agent: Mutt/1.5.21 (2010-09-15)


On Sat, Aug 27, 2011 at 05:47:25PM +0200, Thomas Schmitt wrote:

> my second try ot install of GNU/Hurd on kvm obviously succeeded. :))

Great :-)

> I still have to find out where the RPC from device_get_status()
> arrives and how to perform scsi_ioctl_send_command() on that side of
> the gap.

Indeed, this is pretty tricky, as the device calls go through several
layers of wrappers/emulation...

The implementation of device_get_status() in linux/dev/glue/block.c
seems to be the right starting point -- but I must admit that I'm not
sure where to go from there...

> Even if i had a detailed plan of what to code where, i still lack of
> an idea how the development cycle for a kernel change would look like.

Well, not that much different from userspace stuff: you frob the source,
launch a "make" to build the binary, and then test it...

Which is the part where things are more painful: you obviously have to
reboot the system each time to test your kernel changes. (Or you could
have a second VM just for testing...)

Once you run into problems, debugging is unfortunately also harder, as
the kernel debugger is quite limited. Often the only option is adding
print statements in the code and watching the output on the console...

BTW, do you use IRC? #hurd on freenode should be a good place to get
some handholding :-)

> As userland programmer i feel apt to deal with the requirements of
> users-guide/using_gnuhurd.html#Helping-Out But i never did development
> work inside kernels. In contrary i always took much care not to alter
> the system as side effect of my work. :))

Don't be afraid -- it's really not as scary as most people seem to think
:-) Sometimes kernel development is actually easier, as the kernel is a
pretty self-contained environment, and results of your work generally
very direct --  while in userspace you can have much more complex

> Will a 3 GB "disk" suffice ? Shall i install Hurd to a larger one ?

Well, it should suffice, as long as you only build small packages with
few dependencies. For anything larger, it will be too small very soon --
a glibc build alone for example takes something like 1 GB.

> I will then have more fun with trying to get a real CD drive working
> under control of libburn on a virtual Hurd.

Hehe :-)


reply via email to

[Prev in Thread] Current Thread [Next in Thread]