Re: Interface for SCSI transactions ?

[Samuel Thibault]

Hello,

Thomas Schmitt, le Sat 27 Aug 2011 17:47:25 +0200, a écrit :
> The plan sketched a few months ago was to extend device_get_status()
> for this, because it already provides two-way transmission of an
> integer array. One would have to create a new dev_flavor_t value,
> which indicates that dev_status_t status actually contains a serialized
> struct sdata for parameter buffer of
>   int scsi_ioctl_send_command(Scsi_Device *dev, void *buffer)
> 
> I still have to find out where the RPC from device_get_status()
> arrives

AIUI it's device_get_status() in linux/dev/glue/block.c for the linux
IDE & SCSI devices.

> and how to perform scsi_ioctl_send_command() on that side
> of the gap.

Simply unpack, call, and repack.

However, I realize that the status parameter of the device_get_status
RPC is out only, not inout, so device_get_status alone can not provide
complete support. One way would be to use a set/get pair. Another way
would be to define a new device_command RPC, and extend the
device_emulation_ops structure with it.

> 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.

Simply rebuild gnumach and reboot with it.

Samuel