Opened 4 months ago

Closed 3 weeks ago

Last modified 3 weeks ago

#18949 closed bug (duplicate)

kernel not loading with more than 4GB of RAM

Reported by: zeldakatze Owned by: nobody
Priority: normal Milestone: Unscheduled
Component: System/Kernel Version: R1/beta4
Keywords: Cc:
Blocked By: #14659 Blocking:
Platform: All

Description

Hello, I noticed that on my desktop computer (a Ryzen 5 2600 with an Asus B450+ Mainboard, 16GB of RAM, booting with CSM) Haiku (hrev56578+96) won't boot unless I select the ignore memory above 4GB option. It seems like the kernel won't even run, as there is neither onscreen debug output nor a syslog that is being generated. I've tried updating to the newest nightly aswell (at the time (4th July) beta4_hrev57804), but with that, it won't boot at all, regardless of which boot options I select. I will try to update to the newest nightly aswell. The old state is booting, so I don't think it is a problem with the bootloader. I could also try booting it with UEFI, but currently, I've set it up with legacy CSM to be able to boot without having to redo some partition layout from my preexisting partitions. I'll attach the syslog of the successful boot with hrev56578+96.

kind regards, zeldakatze

Attachments (2)

syslog (164.8 KB ) - added by zeldakatze 4 months ago.
syslog (hrev56578+96, limited to 4gb ram)
previous_syslog (79.6 KB ) - added by zeldakatze 3 weeks ago.
syslog with hrev58228, system was reset shortly after the assertions as to not cut off the debugger stacktraces

Download all attachments as: .zip

Change History (7)

by zeldakatze, 4 months ago

Attachment: syslog added

syslog (hrev56578+96, limited to 4gb ram)

comment:1 by waddlesplash, 4 months ago

Keywords: kernel 4gb ram hangs startup removed
Platform: x86-64All

I'm a bit confused. Is there a version that boots without "4GB memory limit" option enabled? Or does every hrev only boot with that option set?

I think Ryzen systems have always needed this option when booting with BIOS/CSM, see #13370 (which was fixed, but it just uncovered another issue which is also linked in that ticket.)

comment:2 by waddlesplash, 3 weeks ago

Blocked By: 14659 added
Resolution: duplicate
Status: newclosed

The syslog is cut off, but presumably this is another duplicate of #14659. If it isn't fixed in hrev58212, please comment.

comment:3 by zeldakatze, 3 weeks ago

The kernel starts now, though there are multiple Assertions being triggered, mainly about caching. The system starts however, you have to wait for some time until the mouse is actually usable. The system crashed after ~10 minutes whilst checkfs was running with a general protection fault.

I'll attach the syslog, though it seems to me, that the debugger was not logged. I can try to append it's messages aswell

by zeldakatze, 3 weeks ago

Attachment: previous_syslog added

syslog with hrev58228, system was reset shortly after the assertions as to not cut off the debugger stacktraces

comment:4 by zeldakatze, 3 weeks ago

is there a way to increase the syslog buffer size?

comment:5 by waddlesplash, 3 weeks ago

These assertions are the same as #19168.

Note: See TracTickets for help on using tickets.