search

MyersResearchGroup / ATACS

Apache License 2.0
9 stars 1 forks source link

BAD812 broken repository! #37

Open buggsley opened 8 years ago

buggsley commented 8 years ago

Linux ming.ece.utah.edu 2.4.18-19.7.x #1 Thu Dec 12 09:00:42 EST 2002 i686 unknown cputime unlimited filesize unlimited datasize unlimited stacksize 8192 kbytes coredumpsize 0 kbytes memoryuse unlimited descriptors 1024 memorylocked unlimited maxproc 2044 openfiles 1024

CHECK THESE ERRORS FROM MAKE FIRST: vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv make[2]: [.alu1b.prs-tsys] Error 5 make[1]: [csp/check.subDir] Error 2 make: [posets] Error 2 make[3]: [.mergex.merge.prs-oioqtbG1-2ys] Error 1 make[2]: [mergex/merge.VHDLtest] Error 2 make[3]: [.newMergex.newMerge.prs-oioqtbG1-2ys] Error 1 make[2]: [newMergex/newMerge.VHDLtest] Error 2 make[1]: [vhd/VHDLtests.subDir] Error 2 make: *** [channelTests] Error 2 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Messages from log files : vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv ./csp/alu1b.log:Finding redundant rules ... sh: parg: command not found ./csp/alu1b.log:sh: parg: command not found ./csp/alu1b.log:sh: dinotrace: command not found ./csp/alu1b.log:ERROR: Safety violation for rule a1prech+/3 -> a1m-/2. ./csp/alu1b.log:Finding redundant rules ... sh: parg: command not found ./csp/alu1b.log:sh: parg: command not found ./csp/alu1b.log:sh: dinotrace: command not found ./csp/alu1b.log:ERROR: Safety violation for rule a1prech+/3 -> a1m-/2. ./csp/alu1b.log:Finding redundant rules ... sh: parg: command not found ./csp/alu1b.log:sh: parg: command not found ./csp/alu1b.log:sh: dinotrace: command not found ./csp/alu1b.log:sh: parg: command not found ./csp/alu1b.log:sh: parg: command not found ./csp/alu1b.log:sh: dinotrace: command not found ./csp/alu1b.log:ERROR: Safety violation for rule a1prech+/3 -> a1m-/2. ./csp/alu1b.log:ERROR: Safety violation for rule a1prech+/3 -> a1m-/2. ./vhd/atacs.log:ERROR: z_send+/2 enabled but signal is not 0 or R in state 011F10 ./vhd/atacs.log:ERROR: Verification on most-concurrent starting point failed! ./vhd/THE_SHOP.log:ERROR: Safety violation for rule shop1_to_patron_send- -> winery_to_shop1_rcv-. ./vhd/PP.log:ERROR: Safety violation for rule l_rcv- -> r_send-. ./vhd/PA.log:ERROR: Safety violation for rule l_rcv- -> r_send-. ./vhd/AP.log:ERROR: Safety violation for rule l_rcv- -> r_send-. ./vhd/merge.log:System deadlocked in state 100000 ./vhd/merge.log:ERROR: Verification on most-concurrent starting point failed! ./vhd/newMerge.log:ERROR: z_send+/2 enabled but signal is not 0 or R in state 011F10 ./vhd/newMerge.log:ERROR: Verification on most-concurrent starting point failed! ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

buggsley commented 8 years ago

I got a message about the nighttime regression tests last night. (See the above report number).

Based on the tags, it looks like it's related to your checkin with the message "New postprocessing."

buggsley commented 8 years ago

Correction: It was not last night that I got the message, but the night before (early yesterday morning)

buggsley commented 8 years ago

The problem has to do with the function call on line 1340 in postproc.cc to the function clear_C. This function strips "non-essential" conflicts.
These are defined to be if t1 and t2 conflict and don't share a common enabling or enabled event, then they do not form an essential conflict.
In other words, they do not affect behavior.

The problem is that in the hse and CSC solving code, they assume that such conflicts exist. In other words, these two functions modify the tel by adding signals which may be common to these two transitions making this now an essential conflict.

The reason that this function is there is that non-essential conflicts were preventing postprocessing from removing dummy transitions that it can otherwise remove. To make the regression test work, I have commented it out for now, but it will impact postprocessing's performance.

In the future, one of the following needs to be done:

1) Implement unclear_C to put the conflicts back after postprocessing.

2) Implement unclear_C to derive where the non-essential conflicts should be inserted.

3) Fix CSC solver and hse code to not rely on these conflicts.

Options 1 and 2 are the most likely to be done.

Moving to potential problem as this requires future work.