Cursor hangs on startup

[meanwhile]

On startup, Haiku rev. #24385 has a frozen cursor on a blue desktop with nothing on it.

[meanwhile]

I got it to run normally just once, which was after choosing 'Default' Video Mode in the Haiku Boot Loader menu. It's maybe because the screen prefs were adjusted after that, that this couldn't be repeated. Running Haiku in VESA mode doesn't help either.

[mmlr]

This seems pretty obvious to me:

auich: installing interrupt : 9
More than 99% interrupts of vector 9 are unhandled

So from the syslog it starts immediately after auich apparently installed it's interrupt handler (and probably activated the interrupts). Also checking the syslog the only devices present at interrupt line 9 are audio, a modem and SMBus. So it would make sense if this was actually triggered by auich.

[meanwhile]

Thanks for kicking the ball onward, mmlr!

[korli]

Could you try again with r25476+ and provide the usual serial output ?

[meanwhile]

Serial output is in attachment 'serial25477'; the behaviour's unchanged.

[korli]

I need to patch the ac97 code. I hope to do so soon.

[korli]

Could you try again with r25516+ please ?

[meanwhile]

I will. (but there are haven't been any new builds after r25477 from May 12th)

[meanwhile]

Mixed results now (with r25565): the attached serial data show the second boot which was successfull, then the third boot, which wasn't and the fourth, which wasn't either (but got further and showed a Terminal window). I didn't catch data from the first boot as I had the cable connected to the wrong serial port at that time...

[korli]

Could you try again with r25574 please ?

[meanwhile]

It now shows the Deskbar each time, but the boot process ends there; the OS freezes.

[korli]

Once again, could you check with r25592 ?

[meanwhile]

(Please ignore the first attachment, r25593serial.txt. Use r25593serial2.txt instead) The last message from the log seemed to repeat forever, so I shut down the PC after a while. Perhaps because of the size of the log caused by this behaviour, copy and paste didn't save the very first part of the log, or I somehow messed up. Let me know when and if you need this part too. Sorry!

[korli]

Could you check with r25616 ?

[meanwhile]

Hi korli, well it works now: the OS boots nicely (and fully).

BTW, I never had any sound from Haiku yet, but this may be too early, judging by the fact that all sliders in Media prefs. are muted by default and immediately return to that situation after 'unmuting' each and then closing the prefs. window.

[meanwhile]

Thanks for your patience and hard work!

[korli]

Replying to meanwhile:

Hi korli, well it works now: the OS boots nicely (and fully). BTW, I never had any sound from Haiku yet, but this may be too early, judging by the fact that all sliders in Media prefs. are muted by default and immediately return to that situation after 'unmuting' each and then closing the prefs. window.

Didn't understand : there is no sound ? Could you post a serial output ?

[meanwhile]

Earlier Haiku revisions on this PC never had working sound, so in fact this may be a seperate issue/bug. In r245619 I do notice a very short crack through my headphones at the moment after the Terminal shows up and before the Deskbar shows up, so there was an audio signal for a short moment. In 'Media' preferences window, when selecting 'Auich ICH' on the tab called 'AC97 Mixer', all the sliders have 'Mute' selected with a blue cross. When removing the blue crosses and then closing the Media preferences window and re-opening it, the blue crosses are there again; that's what I meant by "all sliders are muted". Anyhow, I hope the r245619boot.txt attachment will be informative.

[korli]

Could you check again ? Rudolf made a change which could help in your case.

[meanwhile]

Replying to korli:

Could you check again ? Rudolf made a change which could help in your case.

It did help...thanks both of you; CD playback is very good and I'll test other audio files. (A 'BeStartup.wav' from BeOS is a bit troublesome for example, but that might end up in another ticket) Thanks again :-) ...and this bug can be closed, I think.

[korli]

Replying to meanwhile:

It did help...thanks both of you; CD playback is very good and I'll test other audio files.

Ok, we would need to check accessed registers, some could be invalid uses.

(A 'BeStartup.wav' from BeOS is a bit troublesome for example, but that might end up in another ticket)

Yes. I have problems with such files too.

Thanks again :-) ...and this bug can be closed, I think.

[korli]

Replying to korli:

Replying to meanwhile:

It did help...thanks both of you; CD playback is very good and I'll test other audio files.

Ok, we would need to check accessed registers, some could be invalid uses.

Meanwhile, could you check again with r25978 and provide the serial output ? Thanks.

[meanwhile]

Hi korli, the attachment is from the 1st boot of a new install (r26003). Cheers, Meanwhile.

[korli]

Replying to meanwhile:

Hi korli, the attachment is from the 1st boot of a new install (r26003). Cheers, Meanwhile.

Thanks. It seems ac97_reg_is_valid() should be updated for CODEC_ID_AD1885, avoiding 0x56.

[korli]

Replying to meanwhile:

Hi korli, the attachment is from the 1st boot of a new install (r26003). Cheers, Meanwhile.

Could you check again with r26168 and provide the serial output ? Thanks.

[meanwhile]

I'm stuck for the moment on another issue that prevents me from running a new revision on this hardware...(see bugs #2461 and #2470)

[meanwhile]

Got a recent working Haiku partition again: r26375

[korli]

Thanks. Bug really fixed.