[BUG] Dell Unix panics while extracting files

[win2kgamer]

In PCem v15 as well as PCem-dev commit 6bd1c70, Dell Unix Release 4 Issue 2.2 panics while extracting files (after the 30 base system floppies are copied). This occurs in both the 486 interpreter and the 486 recompiler:

Screenshot from 2020-03-11 23-03-20.png (7.72 KiB) Viewed 4501 times

In PCem v14, the extraction succeeds and the installer asks if additional packages are to be installed:

Screenshot from 2020-03-11 22-24-10.png (9.28 KiB) Viewed 4501 times

Configuration used on all three versions:
Machine: [486] AMI 486 clone
CPU: i486DX2/66
Memory: 8MB
Video: Tseng ET4000AX
HDD: 250MB IDE
FDD: 1.44MB 3.5"
Mouse: Microsoft serial

EDIT: Issue also occurs when using the AMI 386DX clone with an i386DX/33 and 8MB of RAM. I did four tests with 16MB on the 386 configuration and I couldn't get the issue to occur at all on PCem-dev commit 6e158be. Changing the above 486 configuration to have 16MB of RAM sidesteps this issue as well. Kind of strange that increasing the memory size seems to fix the issue, especially when an 8MB configuration worked in v14.

[win2kgamer]

Currently retesting this OS with commit 6e158be, the kernel panic during decompression seems to be resolved (at least with the recompiler enabled) as I was able to successfully extract the base OS. I am currently installing the additional packages (43 floppies!) and will post an update once the OS install is done.

EDIT 03-22-2020: Issue is still present but occasionally does not occur. See bottom post for more information.

[win2kgamer]

OS install was successful:

Screenshot from 2020-03-21 22-38-25.png (8.32 KiB) Viewed 3822 times

The panic still occurs in both the 486 recompiler and interpreter but is inconsistent.

The panics seem to be inconsistent as I have done multiple tests with both the interpreter and recompiler in v16-dev commit 6e158be and both still experience the panic about 50% of the time. The crash occurs on both SCSI and IDE disks.

[win2kgamer]

The issue seems to be frequent but not super-consistent in v15/v16-dev commits 6bd1c70 and 6e158be as while the kernel panics occur a lot, it will sometimes work and I can complete an install. I have performed some additional tests with v14 (all base system only installs as the issue, when it occurs, always happens during base system decompression): two with the recompiler and two with the interpreter and I cannot get the kernel panic to occur at all. The information from the original post is still valid and the issue is not resolved.

EDIT: Completed second interpreter test

EDIT: Performed some additional tests today using v16-dev commit 6e158be (4 tests with a 16MB 386, 3 tests with a 16MB 486 using the recompiler, two tests with a 16MB 486 using the interpreter and one test with an 8MB 386). I could only get the issue to occur on the 8MB 386.