TF536 - Freeze on Disk (write) access?

[terriblefire]

I am positive - I use the exact same setup/CF card on both the 536 and 330 (I can even use the exact same CPU on both), the 536 will lock up, while the 330 won't. Could it be related to Gayle emulation? (it does take a few hours of archiving/unarchiving or simply lots of disk writing for it to happen)

They both have the same code for Gayle emulation. It could be the interrupt path.. can someone buzz out the IPL lines on the 68000 socket during the freeze?

[go0se]

Here is a capture while the system is locked up during zip compression of a large directory. Clock still running and IPL0 still doing something. IPL1 constant LOW.

Any other signals useful here?

536_WRITE_FREEZE.png (83.37 KiB) Viewed 2978 times

*Edit, Failure file was Addams Family/Disk. 1 Compression had been running for around 40 mins. IDE access light is ON constantly on the SDtoIDE adapter.

[terriblefire]

Here is a capture while the system is locked up during zip compression of a large directory. Clock still running and IPL0 still doing something. IPL1 constant LOW.

Any other signals useful here?

536_WRITE_FREEZE.png

*Edit, Failure file was Addams Family/Disk. 1 Compression had been running for around 40 mins. IDE access light is ON constantly on the SDtoIDE adapter.

IPL0 is glitching... weird.

Can you guys try the test firmware and test disk? it looks like the interrupt is not being disassserted/acknowledged.

[go0se]

I've flashed the tf536r2_ide_test_kit.zip test firmware from here :

https://www.exxosforum.co.uk/forum/viewt ... =30#p44183

And am using the test disk from the zip in the next post tfdisk.zip:

https://www.exxosforum.co.uk/forum/viewt ... =30#p44186

Are those the correct versions?

I'm assuming the IDE cable/adapter/card should be unplugged during these tests, I get fails on some tests with the adapter connected but none when unplugged.

Here are the results with the IDE unplugged:

TESTIDEINT

DSC_0845.resized.JPG (110.79 KiB) Viewed 2947 times

TESTINT2

DSC_0847.resized.JPG (102.25 KiB) Viewed 2947 times

TESTIPL

DSC_0849.resized.JPG (88.56 KiB) Viewed 2947 times

[terriblefire]

Ok so i think we have a clock domain crossing timing issue where something happens to lock up the logic.

I cannot get into testing this right now. Everything is packed away for house move except my space saver A4000 and A1200. (and 2 sets of very expensive keycaps.)

try adding DTACK to that signal trace and then try halving the max transfer rate until it stops happening... if it does.

[go0se]

Excellent, cheers. I'll give that a try this evening.

[Lynxman]

I will try transferring the file with a heatsink on the CPLD. It's the hottest IC on there. Too hot to touch for long o both my T536's.

[Lynxman]

It still hanged.

[richx]

I've also seen the freezing issue randomly with just reading or booting up, LED on CF adapter stays solid.

[alenppc]

I've also seen the freezing issue randomly with just reading or booting up, LED on CF adapter stays solid.

That's caused by something else, probably a loose mechanical connection with the socket or CPU overheating/not stable at the frequency. This bug literally occurs only when writing for an extended period.