Opened 10 years ago

Closed 10 years ago

Last modified 6 years ago

#10717 closed bug (fixed)

KDL trying to mount boot filesystem during bootup

Reported by: dstritt Owned by: waddlesplash
Priority: blocker Milestone: R1/beta1
Component: Drivers/Disk Version: R1/Development
Keywords: Cc:
Blocked By: Blocking:
Platform: x86

Description

Using the nightly hrev47054, and trying any other dating all the way back to before PM merging this happens since I just started to use nightlies, I get a KDL "Divide Error Exception" durring boot on the center icon (disk with leaf), saying it failed to mount boot filesystem. This is from anyboot burned USB, or ISO burned CDR. Safe mode options, several combinations of them, have no effect. There doesn't seem to be a boot log, because the system freezes, and I have to unplug and plug the computer back in. Also tried both USB drives I have, burned from within Windows usimg ImageWriter.

Computer is HP Pavilion a1440n, modified (RAM, New HD, etc), but processor and motherboard are the original.

I am new to reporting boot errors, so if there is any info you need, or something I need to try to get info for you, please ask.

Attachments (2)

2014-04-11 19.10.18.jpg (125.5 KB ) - added by christof 10 years ago.
KDL stack trace
ensure_blocksize.patch (1.3 KB ) - added by waddlesplash 10 years ago.
Patch v2.

Download all attachments as: .zip

Change History (18)

comment:1 by dstritt, 10 years ago

Upgraded motherboard and CPU, so unless someone can reproduce this on their hardware, you might as well close it. All nightlies are working fine now.

comment:2 by dstritt, 10 years ago

OK, it's happening on new hardware, but I narrowed down the cause. My old system had 2 optical drives hooked up, but one was never used, always empty. I successfully installed a recent nightly, but when I boot up with no disc in the drive, I get the same KDL. Insert any disc (not blank) in the drive, and it boots just fine.

comment:3 by anevilyak, 10 years ago

Component: SystemDrivers/Disk

comment:4 by christof, 10 years ago

I have the same problem when trying to boot hrev47105 (x86 GCC 2 Hybrid) anyboot from USB on my desktop with Intel Core i5-4570. The system has a HDD and a DVD drive (no disk inserted).

I will attach a screenshot with a stack trace now. Please let me know if there is any other useful info I could provide.

This looks very similar to #9489.

by christof, 10 years ago

Attachment: 2014-04-11 19.10.18.jpg added

KDL stack trace

comment:5 by korli, 10 years ago

christof, well, the code here expects a non-zero block size. Can you eventually add a check for "partition->block_size != 0"?

comment:6 by TmTFx, 10 years ago

Exactly the same bug for me on a Asus M5A97R2.0 mobo (on a M5A99X EVO works flawlessy without problems even with empty dvd-rom drive)

unplugging the dvd reader the system boots perfectly

tested on hrev 47114

Last edited 10 years ago by TmTFx (previous) (diff)

comment:7 by luroh, 10 years ago

Blocking: 7665 added

comment:8 by waddlesplash, 10 years ago

Component: Drivers/DiskPartitioning Systems/GPT
Milestone: R1R1/alpha5
Owner: changed from nobody to waddlesplash
Priority: normalblocker
Status: newassigned

I'll build an image for testing when I get the time to. Also, there's a forum post here that mentions this bug: https://www.haiku-os.org/community/forum/how_install_pkgman

comment:9 by waddlesplash, 10 years ago

Hm. It appears that this is a duplicate of #9489, although because it has more info, I'll leave both open. (That ticket has some more development discussion...)

The correct way to solve this is to add a check at http://cgit.haiku-os.org/haiku/tree/src/system/kernel/disk_device_manager/KDiskDeviceManager.cpp#n1358 I think. Would that stop the device from being used if it had media (e.g. CD) placed in it (long) after the boot finished?

comment:10 by waddlesplash, 10 years ago

patch: 01

comment:11 by waddlesplash, 10 years ago

Component: Partitioning Systems/GPTDrivers/Disk

I just got this one myself when mounting and then removing a CD in VirtualBox. Working on a patch that actually works...

by waddlesplash, 10 years ago

Attachment: ensure_blocksize.patch added

Patch v2.

comment:12 by waddlesplash, 10 years ago

This fixed the problem for me. See commit message in patch for more info.

comment:13 by axeld, 10 years ago

Patch looks good, just the comment is superfluous - it's perfectly fine to filter out bogus values when they appear. A better comment would explain why such a value gets there in the first place.

I would think it's already a bug that the identify process is started in this case.

comment:14 by jessicah, 10 years ago

Resolution: fixed
Status: assignedclosed

Fixed in hrev47468

comment:15 by pulkomandy, 9 years ago

Milestone: R1/alpha5R1/beta1

comment:16 by waddlesplash, 6 years ago

Blocking: 7665 removed
Note: See TracTickets for help on using tickets.