I took Bill Davidsen out of the list as the email address [EMAIL PROTECTED]
is bad and is bounced since half a year.
>From: "Mike A. Harris" <[EMAIL PROTECTED]>
>>>>> Without ide cdrom support the kernel must add support for an entire
>>>>>SCSI subsystem, which is totally not needed on most systems. I can't see
>>>>>making this a standard, most systems just don't need it.
First as Alan is on the list, I need to answer this again because he used this
argument too:
It is hard to check for the SCSI subsystem size on Linux. On Solaris
it is 30 kB for the SCSI glue layer and about 40 kB for a host adapter
driver. Less than 100 kB of code is no argument for a system with > 64MB
of memory. As most systems installed in this year have 512 MB of memory
it is becoming a joke to use this argument.
Or is the real reason that there are known bugs in the SCSI system on Linux
and you like to avoid to include buggy parts if possible?
>>>>Well most systems now include CD-writers and you need a SCSI subsystem
>>>>to make them work.
>>
>>>While I disagree with that, I must point out most != all. Are
>>>you suggesting that anyone not having a CD writer be abandoned as
>>>far as support is concerned?
>>
>> I see no sense in your mail.
>>
>>It is completely unimportant whether all people have a CD writer!
>I disagree.
So you don't care to get the right solution because you don't understand it.
Please give me a real argument for keeping ide-cdrom in current kernels!
If you disagree, you need to convince me. If you don't give arguments, you
never will convince me.
>>There is a growing majority of people with a CD writer and
>>people like e.g. Alan Cox who don't understand what ATAPI is
>>are going to force these people to live with a badly structured
>>and limited Linux kernel.
>If you're going to mention Alan Cox, I think it is only fair that
>Alan should have a chance to respond to your statements, so I've
>added him to the Cc: list.
Well first let me proove my statment:
About half a year ago, I asked Alan whether he believes if it is possible
to open or close the tray of a CD-ROm drive by only using IDE commands.
As he answered with 'yes', there is a proove that he does not know
how ATAPI works.
... let me explain:
There is a IDE command called 'packet' which is used to send SCSI commands
to the drive using IDE as a transport only.
While ATAPI CD-ROM's support the IDE read command like IDE hard disks,
all other features only work by sending a SCSI command to the drive using
the IDE Packet command. So the right way to support ATAPI drives is to
use the SCSI subsystem to send the commands via a special host adaptor driver
that knows how to send SCSI commands via IDE transport.
If you like to keep ide-cdrom, you need to be realistic and implement a
firewire-cdrom and a USB-cdrom driver too. USB & firewire are just other SCSI
low level transports.
>>- ide cdrom is limited and does not reflect reality
>>
>>- ide-scsi is _not_ limited but implemented buggy on Linux
>>
>>You are saying the same thing as Alan Cox:
>>
>> Let us keep the bugs in ide-scsi & the SCSI cdrom driver on Linux because
>> I do not realize any problem for _my_ personal work.
>>
>>The right solution would be to fix the bugs in the existing ide-scsi + SCSI cdrom
>>solution. Then even people who do not understand the problem would prefer
>>this solutuion because it fits all their needs.
>I haven't said much at all other than that I disagree with you,
>so please do not put words in my mouth like that.
If you disagree without giving any argument, then you are forcing me to
tell you what your ideas finally mean. If you disagree with my statements, then
you finally disagree with yourself.
>I will let Alan comment on the items that you have provided,
>since I know not what his exact words or comments are on this.
I sent Alan commants on the problem many times.
I even wrote him that it is a really bad thing that you cannot write a CD on
a popular notebook as the Sony Transmeta CV-1E
While the CD-writer (Note Sony does not sell CD-ROM's for this notebook anymore)
works as a reader by accident (it is bound as IDE hard disk), you cannot use it
as a writer.
The reason is that you cannot put ide-scsi on top of a IDE tansport that sits on
top of PCATA.
- Linux does not support PCATA with DMA
- Linux crashes or gives funny messages when trying to put ide-scsi on top
of the PCATA transport.
I don't have the time (about 2 days for me to install the latest kernel and
find out what is going on (**1)) to send Alan a better description of what happens.
... but if he is doing serious work (for which I am sure) he definitely has
the hardware to do the same tests in a few minutes.
(**1) On Solaris you have a kernel debugger and crash dumps that may be annalysed
later ooffline. On Linux I see no way to debug the kernel.
J�rg
EMail:[EMAIL PROTECTED] (home) J�rg Schilling D-13353 Berlin
[EMAIL PROTECTED] (uni) If you don't have iso-8859-1
[EMAIL PROTECTED] (work) chars I am J"org Schilling
URL: http://www.fokus.gmd.de/usr/schilling ftp://ftp.fokus.gmd.de/pub/unix
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
