I tried to copy the MorphOS 3.20 ISO (so about 471Mb) on an SFS partition, but when calculating the MD5 of the file, I get random values (both with Ambient information window, and the new command line tool OFHash), almost never the same, and certainly not the correct one.

I guess there's memory corruption happening somewhere, but I don't know where:

- in memory?
- on the hard drive?

That's a Pegasos 2 with the built-in IDE port, and a single stick of 512 Mb RAM. In the serial logs, SmartFirmware says that memory test is OK.

What can I do to locate the problem? Is there a way to do an extensive memory test from MorphOS?

I read about the memtest command, and ran 'memtest EXTENSIVE DISABLE 20 VERBOSE' and got no error.

I noticed this issue a few years ago, on Pegasos 2/G4 too.
The interesting point is that I noticed it with sata while you have it with IDE.
It would confirm it is a generic issue.

Do you have other devices attached to the IDE?
On sata I have an optical drive too.

I have an Harddrive on the first channel, and a Sata SSD with a IDE > SATA adapter on the second channel. Corruption seem to appear on both. I guess I could try booting with one of the channel empty too to see if corruption still happens.

Also, hash is also wrong when reading from USB/FAT16 (which I know for sure has the correct unaltered version written on), and several MD5 also produce different hashes from USB.

So you used a SATA PCI card?

Just wondering: did you have a single or two memory sticks?

Edit: corruption also appears when a single IDE channel is used.

[ Edited by Leo 29.06.2026 - 07:23 ]

Yes PCI sata card; works great but not bootable from SmartFirmware (I have a CF on IDE with just the boot.img).

[ Edité par OffseT 29.06.2026 - 07:50 ]

Does it happen when you do that on files on Ram Disk?

Quote:

This file is too big to fit in RAM: I'll do some tests with smaller files.

OK, I did the test again, and I can still reproduce it.

On my Peg2, it is not that often; I have a wrong md5 only about once every 50 times.
And only on sata hdd; I have no error on Ram Disk.

I could try on IDE later too.

@Leo: I have 2x1GB of RAM (Peg rev 2B5).

Quote:

I did some tests and when running the MD5 from USB or RAM is appears (I say appears because USB 1.1 is so slow I barely have the time to run a single or two checksums before the MorphOS demo limit kicks off) to be OK. But once it's run from the HD, if the file is big enough (like 471Mb for the MorphOS ISO), I get random hashes. Retrying returns another one.

I wish I could try some old Linux distro but unfortunately I wasn't able to find one that can be installed without a CDRom drive.

Could it be an issue of the IDE cable? It has to be a 80pin UDMA cable, a 40pin IDE cable won't work IIRC.

Quote:

The issue is not related to IDE, SATA is also affected.