Opened 5 years ago

Closed 5 years ago

Last modified 5 years ago

#16168 closed bug (duplicate)

Paladin crashes into KDL

Reported by: roiredxsoto Owned by: nobody
Priority: high Milestone: Unscheduled
Component: File Systems/FAT Version: R1/Development
Keywords: Cc:
Blocked By: #7121 Blocking:
Platform: All

Description

Good day,

Haiku installed on real hardware.

Expected behavior:

  • Launch Paladin
  • Work on a Project

Actual behavior:

  • Launch Paladin
  • Jump to KDL, need to reboot ( 'save-report' not working in KDL)

This started to happen just a moment ago, with the last hrev54278.

Attached is syslog

Attachments (3)

syslog (83.7 KB ) - added by roiredxsoto 5 years ago.
Syslog
syslog.old (512.0 KB ) - added by roiredxsoto 5 years ago.
20200612-Syslog.old
Paladin-LeCinquiemeElement.jpg (4.0 MB ) - added by roiredxsoto 5 years ago.
Picture of KDL. Launch Paladin and this is what is gotten

Change History (15)

by roiredxsoto, 5 years ago

Attachment: syslog added

Syslog

comment:1 by X512, 5 years ago

Can you take photo of KDL?

comment:2 by X512, 5 years ago

hrev54278

Can be related to mmlr changes. Something can be broken because all patches was not applied.

in reply to:  2 comment:3 by mmlr, 5 years ago

I can't seem to reproduce this. Can you share more details? Is this on x86_gcc2 or on x86_64 and are you opening a specific project or just launching the app?

Replying to X512:

hrev54278

Can be related to mmlr changes. Something can be broken because all patches was not applied.

The changes are all standalone, I tested them individually. The ones that weren't applied are related to MADV_FREE and some cleanups, they shouldn't matter in this context.

comment:4 by mmlr, 5 years ago

I can't reproduce on a freshly downloaded nightly of hrev54278 x86_64 either.

From the syslog I see that it's x86_64. The syslog doesn't include a full boot and it doesn't contain information about any KDLs (fatal ones cannot be written back to the file in any case, they can only end up there if you can exit from KDL and the system is in a state where it can still write). So the syslog here is of limited use, maybe syslog.old contains more.

It does however state:

KERN: bfs: Oh no, value 3146359 not found in fragments of node 221184..., inode 5479

That doesn't sound healthy. It looks like there's a problem with some of the BFS structures, maybe the KDL is related to that? You can try running checkfs on the volume, but be aware that it may remove things it cannot open (for any reason, actual corruption or just by running out of memory), so make sure to back anything up that you still need.

by roiredxsoto, 5 years ago

Attachment: syslog.old added

20200612-Syslog.old

comment:5 by roiredxsoto, 5 years ago

Good day, Added a syslog.old of 20200612, this syslog.old was produced just right now, so it's fresh. Just launching Paladin produces the trip to KDL, while before issue appeared when picking a project. And there is no response from KDL yet; I mean, I can type, nothing appears on the screen, and in the end still can't save a report. I also took a pic of the KDL that I will post later as it's on the phone.

It did work without issues before. Which makes it weirder.

The system is installed on an USB3 pendrive, 128 GB, in UEFI mode and runs on a Ryzen 5 1600 with 24 GB RAM. I presume memory is not an issue here. I did deactivate Virtual Memory and ran the checkfs on the System partition after doing so (I do this -virtual memory- the moment I install Haiku on anything).

I will do the checkfs and report any issues later. I don't recall any issue when I did it when deactivating the Virtual Memory though.

Whatever info you need let me know, I will try to provide it as soon as I can.

Thanks. Regards, RR

comment:6 by roiredxsoto, 5 years ago

This is the outcome of "checkfs /Haiku":

        6526 nodes checked,
        0 blocks not allocated,
        0 blocks already set,
        0 blocks could be freed

        files           4091
        directories     2159
        attributes      160
        attr. dirs      107
        indices         9

        direct block runs               6453 (3.07 GiB)
        indirect block runs             3 (in 1 array blocks, 228.19 MiB)
        double indirect block runs      0 (in 0 array blocks, 0 bytes)

Seems fine, right?

comment:7 by X512, 5 years ago

I also took a pic of the KDL that I will post later as it's on the phone.

It is hard to understand what is happening without KDL photo. It should contain error message and stack trace.

by roiredxsoto, 5 years ago

Picture of KDL. Launch Paladin and this is what is gotten

in reply to:  7 comment:8 by roiredxsoto, 5 years ago

Replying to X512:

I also took a pic of the KDL that I will post later as it's on the phone.

It is hard to understand what is happening without KDL photo. It should contain error message and stack trace.

Added the pic, X512, if you need I carry some test procedure in order to gather more info, just let me know. Thanks. Regards, RR

comment:9 by X512, 5 years ago

Problem seems to be in FAT file system. Can you check FAT volume in Windows?

in reply to:  9 comment:11 by roiredxsoto, 5 years ago

Replying to X512:

Problem seems to be in FAT file system. Can you check FAT volume in Windows?

Ok. will do and report. FAT file system is a "shared" drive and didn't present any issue until lately. It's true that I had issues with the partition as before was NTFS, then changed to FAT. Maybe issue is related to that?

Edited:

  • I just wanted to do some test before booting windows and check the filesystem. I unmounted the FAT volume and launch Paladin. This time, everything went fine. Paladin launched without issues, can create new projects or open those that I'm working on. So certainly it was related to the mounted FAT partition.
  • I will check the filesystem in Windows and then report properly.

Thanks again X512! Regards, RR

Last edited 5 years ago by roiredxsoto (previous) (diff)

comment:12 by diver, 5 years ago

Blocked By: 7121 added
Component: ApplicationsFile Systems/FAT
Keywords: Paladin Crash KDL removed
Resolution: duplicate
Status: newclosed
Note: See TracTickets for help on using tickets.