freenode/#mezzano - IRC Chatlog

16:41:04 fittestbits__ OK, I'll make the eval->compile change and type-of -> array-element changes, I'll save them up for a single PR comments commit.

16:53:33 fittestbits__ I'll see what I can do with the system name ...

19:12:11 froggey fittestbits__: I'm getting a crash during probe (big crash, vbox gives up with a guru meditation error)

19:13:09 froggey I think it's because you're doing 16-bit accesses to the registers with pci-io-region/16, the spec says accesses must be done with 32-bit accesses

19:14:57 froggey specifically the crash occurs when +command-controller-reset+ is written to +ohci-command-status+, exactly on the write instruction & with a vbox error that looks like it doesn't know how to handle the access

19:17:02 froggey VirtualBox VM 5.2.18_Ubuntu r123745

19:43:58 froggey_ ** NICK froggey

19:45:23 froggey switching the pci accesses over to /32 seems to fix it

19:50:52 random-nickname ** NICK random-nick

19:53:44 froggey passthrough is working with my mouse

19:53:58 froggey though I got an error the first time I tried: https://gist.github.com/froggey/b083976cf5c2bc43f5539d596bcfea77

19:54:56 fittestbits__ Wow - OK I'll look at the /16 thing. I use it in a number of places. I'll take a look at the OHCI spec as well.

19:55:08 fittestbits__ It does seem wrong to me that I'm using /16 instead of /32.

19:57:47 fittestbits__ FYI: I'm using Vbox 6.0.10_RPMFusion r132055 on Fedora 29.

20:05:39 fittestbits__ I don't know why there are /16 references, all of the registers are 32-bit registers. Who wrote that code??

20:14:12 froggey it doesn't seem to detect devices that're already connected on boot. I have to disconnect/reconnect to trigger a rescan, then everything gets picked up

20:15:58 fittestbits__ Are you configuring VBox so that it always picks up the USB mouse? Rather than having to go through the menus after boot?

20:16:08 froggey yes

20:17:06 fittestbits__ OK - I didn't think of trying that - sorry. I'll work on it.

20:19:26 froggey that's how virtualbox's usb tablet works too

20:19:41 froggey aside from those issues it all looks really good

20:19:48 fittestbits__ I'm looking at your gist now ... not sure what went wrong. I'd like to understand what went wrong, not just put in a check for NIL.

20:20:34 fittestbits__ Thanks

20:22:05 froggey I haven't been able to reproduce it and I didn't see any more information in the log

20:23:49 froggey it might have happened because I added the device partway through boot, maybe the controller was only half-configured at the time

20:23:58 froggey but don't read too much into that, I might have actually clicked the button after the system had finished booting. I'm not really sure

20:31:43 fittestbits__ Oh, I wonder if I'm missing a with-hcd-access somewhere ... or a race condition in tear down.

20:46:10 fittestbits__ froggey: What did you have in mind for a better name for the system: drivers-usb? lispos-usb? The file name and system name should match for asdf searching to work ...

20:46:30 fittestbits__ Any name is fine with me.

20:52:18 froggey I didn't think that far ahead. I guess mezzano-usb? to indicate that it's mezzano-specific

20:56:55 froggey I can reproduce various errors by attaching a device late during boot, after DHCP lease acquisition or shortly after the old controller is deleted

20:57:57 froggey https://gist.github.com/froggey/b083976cf5c2bc43f5539d596bcfea77#file-gistfile2-txt

20:58:14 froggey I turned ohci's *trace* variable up to 1000

21:06:10 fittestbits__ Thanks for the log. I took a quick look at finding devices on boot. I think it's going to take a while to work through supporting that.

21:06:25 fittestbits__ Also I'll take a look at the log and see what needs to be fixed there.

21:06:53 fittestbits__ So, it may be a while before I have updates for these problems.

21:11:10 froggey alright