I reverted back that code I altered, I'll leave it to someone knowledgeable.

Meanwhile I had to relax timeouts in ahci code as when with no debug output Aros installer always failed at the unicode file (something, something, forgot the name but begins with _) while copying grub files.

I changed the timeout from 1000ms to 1500ms and it no longer chokes on that file.

I don't know why it always failed on that one single file, it maybe because of the drive or the cd-rw disc used (TDK 4x-12x, 80min 700mb)

Docking station arrived yesterday, unfortunately USB3.0 ports are implemented as a superspeed hub and it only passes through the usb lines to the laptop and in my case there is no xHCI controller on Dell Latitude E6410 -> ports are USB2.0 only...

Good news is that Aros is able to use the serial port for debugging.

But the main issue is that ahci.device does not recognise the presence of a disk (cd/dvd) correctly and by looking at the source it is clear why that is. Or is it that clear...

Explicit sense request are disabled/not implemented and SCSI_TEST_UNIT_READY always return hard coded success. SCSI_TEST_UNIT_READY is used in the monitor task that should tell if the disk presence has changed.

This is not really my cup of tea, better revert my commit.

I'll wait for the docking station, it will then be a lot easier to see what is happening with my drive.

Should I report your changes for "Week in AROS" as improving the situation (even if a bit) or skip reporting them at the moment?

While reading on some SCSI-2 manuals it is clear that the commit I made is nowhere near enough or correct. It will take a few days but sometime next week I'll have a docking station for this Dell laptop. It has ordinary serial port for debugging and some USB3.0 ports :)

Never mind, I had a whack at it myself. Some of the issues still remain, for example one can't eject the cd/dvd (at least when it is the boot device) as Aros does not recognize it again. Same happened with the previous version also.

Usb massstorage class has also a bug decoding the sense data, but I do not have a usb cd/dvd drive so I won't touch that.

When CD/DVD is spinned down the drive gets unmounted(or TD_ADDCHANGEINT gets called)

In ahci_cam_aros.c code checks (ahci_PortMonitor) if the device is ready using scsi commands but I assume that when the drive gets spinned down the flag will be set. Spinned down drive is ofcourse not ready but that does not mean the medium has changed.

There is a note about querying sense data (TODO) in ahci_cam_aros.c

USB mass storage code uses the sense data with hardcoded definitions, but the definitions for sense data can be found in ahci_scsi.h

Could someone more knowledgeable take a look? It takes all the pleasure of using Aros when the drive keeps unmounting and mounting.