EISA 53c825 - where to start?

EISA 53c825 - where to start?

[Meelis Roos]

Hi,

I have a Compaq Proliant 4500 with onboard EISA-connected NCR 53c825 
SCSI adapter. There is no EISA-aware version for this chip in Linux 
AFAIK so I want to make one. EISA API is there, also several NCR/Sym/LSI 
SCSI drivers. Which driver is the best to base the eisa work on? And 
where should I start? :)

-- 
Meelis Roos (mroos@linux.ee)

Re: EISA 53c825 - where to start?

[Matthew Wilcox]

On Tue, Nov 16, 2004 at 11:12:20AM +0200, Meelis Roos wrote:
> I have a Compaq Proliant 4500 with onboard EISA-connected NCR 53c825 
> SCSI adapter. There is no EISA-aware version for this chip in Linux 
> AFAIK so I want to make one. EISA API is there, also several NCR/Sym/LSI 
> SCSI drivers. Which driver is the best to base the eisa work on? And 
> where should I start? :)

As you note, we have several drivers for NCR/Symbios/LSI chips.  Here's
a taxonomy:

53c700		Handles 700 chips and 710 chips in 700 compat mode
lasi700		Used with 53c700 for 700/710 chips in PA-RISC workstations
NCR_D700	Used with 53c700 for 700 chips on an MCA bus
sim710		Used with 53c700 for 700/710 chips on MCA or EISA bus

53c7xx		Handles 710 chips on m68k
amiga7xx	Used with 53c7xx for Amigas
bvme6000	Used with 53c7xx for an m68k-based VME machine
mvme16x		Used with 53c7xx for an m68k-based VME machine

ncr53c8xx	Handles 720 chips, should be extended to handle 770 chips too
NCR_Q720	Used with ncr53c8xx for 720 chips on an MCA bus
zalon		Used with ncr53c8xx for 720 chips on PA-RISC machines

sym53c8xx_2	Drives 8xx and 1010 PCI chips

The 5380, 53c9x, 53c416 and 53c500 are unrelated chips.

(it'd probably be a good idea to put this summary somewhere in the kernel tree)

I think your best bet would be to add EISA support to the sym2 driver.
This may turn into a lot of restructuring, so do feel free to ping me
frequently with questions about how I'd like to see things work.

The only other reasonable alternatives are to start a new driver (blech)
or adapt ncr53c8xx to handle EISA 825 ... and I think that's more work,
as well as being not where we want the ncr family to end up.

-- 
"Next the statesmen will invent cheap lies, putting the blame upon 
the nation that is attacked, and every man will be glad of those
conscience-soothing falsities, and will diligently study them, and refuse
to examine any refutations of them; and thus he will by and by convince 
himself that the war is just, and will thank God for the better sleep 
he enjoys after this process of grotesque self-deception." -- Mark Twain

Re: EISA 53c825 - where to start?

[James Bottomley]

On Tue, 2004-11-16 at 03:12, Meelis Roos wrote:
> I have a Compaq Proliant 4500 with onboard EISA-connected NCR 53c825 
> SCSI adapter. There is no EISA-aware version for this chip in Linux 
> AFAIK so I want to make one. EISA API is there, also several NCR/Sym/LSI 
> SCSI drivers. Which driver is the best to base the eisa work on? And 
> where should I start? :)

Are you sure it's an 825?  The reason I ask is because the 825 chip is
both a SCSI chip and a PCI chip with on-board BIOS.  The only real way I
could see it working as EISA is if it were glued to a EISA<->PCI
converter, which I didn't think existed.

Also, the only way to get this to work will be to find out what the
magic glue registers are and how they're mapped ... unless you have
access to the HW specs for the box, this is going to be tricky to
reverse engineer.

James

Re: EISA 53c825 - where to start?

[Meelis Roos]

> Are you sure it's an 825?  The reason I ask is because the 825 chip is
> both a SCSI chip and a PCI chip with on-board BIOS.  The only real way I
> could see it working as EISA is if it were glued to a EISA<->PCI
> converter, which I didn't think existed.

Well, the chip has
NCR
53C825
609-0391625
DP03537
9515N
written on it.

> Also, the only way to get this to work will be to find out what the
> magic glue registers are and how they're mapped ... unless you have
> access to the HW specs for the box, this is going to be tricky to
> reverse engineer.

Hmm. At least Windows and Solaris drivers should work, should be able to 
at least disassmble them if there is no other way - if I ever get that 
far.

A quick glance at NetBSD and FreeBSD didn't reveal a EISA version of the 
driver either :(

-- 
Meelis Roos (mroos@linux.ee)

Re: EISA 53c825 - where to start?

[Marc Zyngier]

>>>>> "Meelis" == Meelis Roos <mroos@linux.ee> writes:

>> Are you sure it's an 825?  The reason I ask is because the 825 chip is
>> both a SCSI chip and a PCI chip with on-board BIOS.  The only real way I
>> could see it working as EISA is if it were glued to a EISA<->PCI
>> converter, which I didn't think existed.

Meelis> Well, the chip has
Meelis> NCR
Meelis> 53C825

There is probably a converter chip somewhere near the 825 (look for a
PLX chip with EISA written on it...)

	M.
-- 
And if you don't know where you're going, any road will take you there...

Re: EISA 53c825 - where to start?

[Meelis Roos]

> There is probably a converter chip somewhere near the 825 (look for a
> PLX chip with EISA written on it...)

No PLX but there is a Motorola chip near it with
SC02RH004DK08
142203-002

And another similar Motorola a bit further:
SC38SG003CN04
127053-002

Google finds both bot I see no realtion with neither pci nor eisa.

There is also a NCR 0390784 chip ner but this seems to be another scsi 
controller for cdrom and tape.

-- 
Meelis Roos (mroos@linux.ee)

Re: EISA 53c825 - where to start?

[James Bottomley]

On Tue, 2004-11-16 at 12:07, Meelis Roos wrote:
> No PLX but there is a Motorola chip near it with
> SC02RH004DK08
> 142203-002
> 
> And another similar Motorola a bit further:
> SC38SG003CN04
> 127053-002
> 
> Google finds both bot I see no realtion with neither pci nor eisa.
> 
> There is also a NCR 0390784 chip ner but this seems to be another scsi 
> controller for cdrom and tape.

Well, the good news is that if you can find this chip and its
documentation then you can simply construct it as an EISA<->PCI bridge
and then the sym_2 driver should work with the 825 chip without
modification.  The Bad news, of course, is that you have to find the
chip and its documentation...

James