Bootscreen stops at 4th icon on Abit IP35

[biffuz]

Hi,

I replaced my ASRock with an Abit IP35, and when I try to boot Haiku (r24870 downloaded from Build Factory), it stops at the fourth icon. However, I can go into KDL and get a stack crawl. This board has not a serial port, so I had to get a "screenshot" :-)

[axeld]

And where is that? Does it end up in KDL by itself, or do you have to press F12 in order to get there? In the former case, a stack crawl would be very helpful, in the latter it would be useless.

If it's in fact the latter, please enable "on screen debug output" in the safe boot menu, and make a shot of the last screen you see (you'll have to press a key after each screen to continue).

Do you have a SATA disk, BTW?

[biffuz]

sc on KDL

[biffuz]

Replying to axeld:

And where is that? Does it end up in KDL by itself, or do you have to press F12 in order to get there? In the former case, a stack crawl would be very helpful, in the latter it would be useless.

I can press F12. Well, I attached it anyway :-)

If it's in fact the latter, please enable "on screen debug output" in the safe boot menu, and make a shot of the last screen you see (you'll have to press a key after each screen to continue).

Why didn't I thought of that? I'll do asap.

Do you have a SATA disk, BTW?

The disk is IDE, connected to (of course) the one and only IDE port of this board. There's a DVD-ROM as slave. I also coonected an IDE DVD-RW to the first SATA port with an adapter, do you think I should try to detach it?

[biffuz]

I tried with on screen debug output, and I get several pages of an error, you can see it in the syslog I'm going to attach. At last page, it says:

IDE: scan_device_int: waiting 100ms... More than 99% interrupts of vector 10 are hunandled Last message repeated 378 times. ide: ide_timeout() bus 90d5e700 More than 99% interrupts of vector 10 are hunandled

and here it stops.

Both errors are relative to the SATA controllers, so I tried to detach the DVD-RW on the SATA port, and booting went fine! I picked the syslog, so you can see at least the first of the errors.

[biffuz]

syslog after detaching SATA DVD-RW

[umccullough]

Disclaimer: I have different hardware, but also experiencing the same problems now as well.

I recently started running into the same thing on my Dell Optiplex GX270 machine (which previously worked perfectly).

I started having the problem when I changed from using a 40gb IDE (UDMA PATA) disk for Haiku to using a 80gb SATA disk instead. (no SATA disks were in use on this machine previously).

In my case, the CDROM is still connected to the PATA controller.

My on-screen debug output looks very similar referencing the "99% interrupts unhandled".

I haven't tried the newer ata driver vs the ide driver yet - so I haven't logged a ticket. Is there a proper way to debug this otherwise?

[stippi]

Based on this info, I am reassigning to Marcus. Could be wrong though.

[umccullough]

FWIW, I replaced the ide bus_manager with the newer ata version and that seemed to clear it up. Of course, I also update to the latest rev, so I will switch back and make sure it still fails with the older "ide" version.

Also, I thought it was still broken as it hangs on the 4th icon for 10-15 seconds before continuing - and definitely seems slower than it was with the PATA disk.

Anyhow, probably going to switch back to the PATA disk in the next couple days until the situation improves, unless there is anything further I can test and provide now.

[marcusoverhagen]

The screenshot shows that ide bus_manager gets stuck in scan_device_int. The ide bus_manager triggers an interrupt but never acknowledges it.

This bug is fixed in the ata bus_manager, but as it's still unfinished, is has a few other bugs, for example a timeout problem when accessing ATAPI devices. I plan to fix those pretty soon and switch Haiku to the ata bus_manager.

[scottmc]

This sounds a lot like #1444.

[richienyhus]

With Haiku now using the ata bus_manager, shouldn't this ticket be fixed?

[biffuz]

I still own that board, so I can give it a try.