[announce] plscsi in Linux

[announce] plscsi in Linux

[Pat LaVarre]

> Re: [PATCH] atapi request sense work
> ...
> 'plscsi'.  Where would one find this tool?

http://members.aol.com/plscsi/linux/plscsi.tar.gz

ends the depressingly noisy web trail:

http://www.google.com/search?btnI=lucky&q=plscsi
http://members.aol.com/plscsi/
http://members.aol.com/plscsi/linux/

There you will find,

plscsi (Programming Language, SCSI) is a tool, conceptually trivial,
actually complex, typed by me (Pat LaVarre) with help from many, but not
yet offered among the sg utils of linux-scsi/ Douglas Gilbert.

I wrote plscsi to make { open(); ioctl(); close(); exit(); }
interactive, taking the SCSI CDB in hex from the command line and
streaming the binary data in/ out thru stdout or stdin.  Relevant ioctl
in Linux include CDROM_SEND_PACKET and SG_IO.

Such an interactive tweak turns out to be exactly what you need, for
example, to make linux-2.6.6-bk8 SATAPI read/write.

A tutorial is http://members.aol.com/plscsi/tutorial/

Linux-style source is http://members.aol.com/plscsi/mac/gcsmac.tar.gz

http://plavarre.blog-city.com/read/542134.htm
is me arguing that a C interpreter would do a better job.

Among the folk of linux-ide I trust we have no need of the disclaimer:

Learning to use a tool like PLScsi will probably remind you of the need
to backup often. If you won't enjoy that lesson, then please don't pick
up the tool. Potential side effects include kernel panics, destroyed
discs, destroyed devices, and even imperceptibly subtle forms of data
corruption. Storage devices differ from modems first by being designed
to record your mistakes permanently, and second by defining some
commands or sequences of commands to mean self-destruct.

Pat LaVarre

Re: [announce] plscsi in Linux

[Pat LaVarre]

> > 'plscsi'.  Where would one find this tool?
> 
> http://members.aol.com/plscsi/linux/plscsi.tar.gz

A noisy list of known bugs in plscsi is:

Subject: plscsi issues
http://plavarre.blog-city.com/read/579210.htm

Without the noise, that signal for Linux today includes:

1) FIXME: Parse the device name to default to O_NONBLOCK for /dev/scd$v
/dev/hd$v despite defaulting to O_RDWR | O_NONBLOCK for /dev/sg$n.

2) FIXME: Search more than just /dev/sg* to build the `plscsi -w` list
of connected devices.

3) FIXME: Add ioctl CDROM_SEND_PACKET to coexist with ioctl SG_IO,
mostly for the sake of ide-cd DVD/CD in Linux 2.4.

4) FIXME: Port back to Linux the master source last released as
http://members.aol.com/plscsi/mac/gcsmac.tar.gz

5) FIXME: Rename the .cpp to .c by removing the inline declarations from
all the source except the Windows port.

Pat LaVarre

P.S. All the noise I've collected on SCSI pass through might be
reachable thru:

Subject: SCSI pass thru
http://plavarre.blog-city.com/read/614076.htm

Re: [announce] plscsi in Linux

[Pat LaVarre]

Someone once said if you ask me the time I will first help you build a
clock and then I may remember to tell you the time.

> .. conceptually trivial, actually complex,
> ...
> http://plavarre.blog-city.com/read/542134.htm
> is me arguing that a C interpreter would do a better job ...
> ... 
> A tutorial is http://members.aol.com/plscsi/tutorial/
> ...
> 4) FIXME: Port back to Linux the master source last released as
> http://members.aol.com/plscsi/mac/gcsmac.tar.gz

plscsi development most recently stalled for lack of conversation over
how to express the C-struct-like idea of consciously misaligning struct/
cdb/ sense/ data.

As yet, the master Mac source is good only for writing short Mac OS X
demoes in the style of http://members.aol.com/plscsi/tools/gccscsi/

The Mac PLScsi source is NOT yet merged with the command line parser.

> Subject: pldd issues
> http://plavarre.blog-city.com/read/596492.htm
> Subject: gccscsi issues
> http://plavarre.blog-city.com/read/617727.htm

The Mac PLScsi source is NOT yet ported back to Linux, even though the o
0 not implied after y/ n/ yn bug of Linux gccscsi is fixed in the Mac
source.

Pat LaVarre