problems in device_manager

[korli]

When running rev16975 image in VirtualPC, I get the following stackcrawl.

[korli]

The weird thing is : it happens when Haiku is on secondary slave. It doesn't happen when Haiku is on primary master or primary slave or secondary master.

[axeld]

* Bug 389 has been marked as a duplicate of this bug. *

[axeld]

I've now reverted hrev16946, does this help the issue? This seems to be a bug in the device manager, that I cannot reproduce.

[korli]

I didn't test your change yet, but I'm wondering why the ide_isa module tries to publish its channels. The code

if (pnp->acquire_io_resources(resources, resource_handles) != B_OK) return B_OK;

in ide_isa:probe_channel() should return and nothing should get published since the pci ide driver handles the ide channels.

[axeld]

Maybe it's first to acquire the resources, maybe there is some kind of race condition. In any way, it shouldn't crash either :-)

Feel free to investigate into this problem, that would be very much appreciated, since I cannot reproduce it :-/

[korli]

ide_isa publish_channel

[korli]

serial log

[korli]

Your revert to hrev16946 seems to have made the bug disappeared. It could also be a race condition indeed. I added some debug output in ide_adapter to know if it fails because of resource acquirement, and which one.

Anyway the crash is gone.

[axeld]

Changed summary and opened it again, since the actual problem remains unsolved.

[umccullough@…]

I also will give this a try tonight to see if it resolves the issue I was having stated in #389

[umccullough@…]

Yes, reverting the change from hrev16746 has resolved the crashing problem here also.

[marcusoverhagen]

did hrev17619 change the behavior?

[axeld]

Outdated.