Slow REL file operations on real hardware

[fachat]

On real hardware REL file operations so slow that the IEEE488 timeout needs to be disabled to actually make them work.
Observed by @snhirsch on PetSD
Originally reported in #177 but separated out here for clarity and tracking

[fachat]

I have a setup with a PET, an early petSD2, and an SFD1001. I tried some tests creating and reading REL files, but never needed to set the timeout to make the REL file operations work.

Do you have a logic analyzed? Can you trace the IEEE488 bus (200kHz sample freq usually works) while such a timeout happens and send me the files (separately)?

[fachat]

Did you actually set the fuses right for XD2031 on the petSD? In the firmware/petSD there is a README that hints on data loss on the IEC bus ... not sure if it may affect IEEE488, I never had a petSD.
I will have a further look at the hardware differences and what they mean for the firmware.

[snhirsch]

I had sent you such a trace last fall :-). Let me do some digging and see if I can find the data.

[fachat]

sorry that I forgot about that. Anyway, did you check the fuses?

[snhirsch]

That's a sore subject. I bricked one MPU already by fiddling with fuses and would rather not repeat this exercise. What exact setting(s) are you referring to? Wait, sorry. I would certainly have run the petSD_fix_fuses.sh script when first installing the firmware. Didn't think it would work unless that had been done.

[fachat]

I am talking about this README https://github.com/fachat/XD2031/blob/master/firmware/petSD/README - must be by @nils-eilers as I've never had a petSD.

[snhirsch]

That would have been done, yes.

[fachat]

just to be sure - you are using this device? http://petsd.net/orgpetsd.php?lang=en

[snhirsch]

Yes, that's it. The only difference is that mine lacks clock and ethernet.

[fachat]

hm. I have an intermediate version "petSD2" between that and petSD+ that lacks Ethernet but has a clock (at least it has a coin cell)....... can you pls verify that pin 1 of the 75160 or 75161 (TE signal) is connected to pin 1 (PB0) of the Atmega 1284? (or even check all signals as defined in firmware/petSD/hwdefines.h?

[fachat]

Also, does the problem also appear with only the petSD attached to the PET (and the other drive(s) completely disconnected, not just switched off)?

[fachat]

There are two operations / test cases that usually test the IEEE488 communication pretty well: using the DIRECTORY (or CATALOG) command on BASIC4, as this basically does a TALK/SECTALK/GET/UNTALK for every byte it reads - and a small test program I have written that dumps some data with single-character PRINT#. As DIR seems to work (or you'd probably have reported it already), can you also check and run this program https://github.com/fachat/XD2031/blob/180-rel-len/tests/seq.lst an see if the result is an 8k file that has just counting bytes like 00 01 02 ... fe ff 00 01 ... in it, and no byte missing?

[snhirsch]

That test passes without incident. All 256-byte chunks are identical.

Here's the schematic for my PetSD:

petSD_schematic_0.7.2.1.pdf