IDE programming documents request ...

[Pype.Clicker]

i need more info about IDE programming, especially the content of the ATA_IDENTIFY returned pseudo-sector.

I've been through http://www.nondot.org/sabre/os/files/Disk/IDE-tech.html, but it doesn't seem to explain everything.
For instance, where is explained that byte order from "model" string must be reversed ? and how can one make sure the HDD will support LBA ?

If some of you has extended knowledge to share ...

[distantvoices]

merde, que j'ai oublie l'adresse excacte d'une website sur ATA ...

alors ...

A quick google search didn't throw it out anyway, so I'll search tomorrow. but I know there is a kind of ATA ressource center lingering around in internet. They offer drivers and source code and ressources ... Crap that I didn't bookmark it ... shame on me and glowing coals amongst my hair ...

[FlashBurn]

Maybe you have a look at http://www.t13.org.

[Pype.Clicker]

hmmm ... seems to be a promizing link.
I'd like to apologize for my laziness ... my modem seems to have nervous breakdown these last days and googling is especially boring under these conditions ... thanks alot.

[Pype.Clicker]

eeek!?
I've got a strange behaviour while trying to detect IDE controllers. The code i use looks like

Code: Select all

   byte r=inb(ATA_CYLINDER_LOW);
   outb(ATA_CYLINDER_LOW,~r);
   if (r==inb(ATA_CYLINDER_LOW) error("controller not found");

It has been borrowed to Tim's ata.c driver and expects that if a controller is present, then we should be able to read back a value we stored in registers like CYLINDER_LOW (that is supposed to be a place where we store info for a read/write request on the disk before the command is issued) but that the port will be read as a constant if there is no controller.

That code worked fine on bochs-1.4.x, but it fails on bochs 2.0.2 (it reads '0' both before and after the 'outb' command ?)

Do anybody experienced such a weird behaviour ?
Is the controller test really useful or is it enough to send an ATA_IDENTIFY command and see if a result is returned ?

[Therx]

Maybe you just typed it wrong but in the last line the if statement is missing the closing bracket.

[Tim]

That method of detection in my code is broken, at least on Bochs. I rely on sending IDENTIFY and waiting for a timeout, although that's obviously very slow. I haven't got round to finding a better solution yet, so maybe it's time one of us did.

[Pype.Clicker]

hmm ... some code extract from BOCHS 2.0.2

Code: Select all

    case 0x04: // cylinder low 0x1f4
      value8 = (!BX_SELECTED_IS_PRESENT(channel)) ? 0 : (BX_SELECTED_CONTROLLER(
channel).cylinder_no & 0x00ff);
      goto return_value8;

Which seems to show that i should have selected the proper drive before starting to talk with the controller ...
BUT if i do select master drive and that it is not present (for instance, what if the IDE bus only has a slave CDROM, but no master ?) how will real hardware react ??
Are the CYLINDER_LOW, etc. registers buffered at the chipset, or do they directly alter the state of the disk's electronic ?

Maybe a valid test would be :

Code: Select all

   atactrl::reset();
   atactrl::select_drive(0);
   if (atactrl::test()) goto found;
   atactrl::select_drive(1);
   if (atactrl::test()) goto found;
   error("controller not found");
   return false;
found:
   // start identifying drives

where "atactrl::test()" performs the CYLINDER_LOW test and "atactrl::select_drive()" issues a outb to set/reset the drive select bit...

[srg]

try this:

http://www.ata-atapi.com/drvr.htm