ABC: execution of command failed return code 139.

[kareefardi]

Version

Yosys 0.42+40 (git sha1 a739e21, g++ 11.4.0-1ubuntu1~22.04 -fPIC -Os)

On which OS did this happen?

Linux

Reproduction Steps

test-case.zip
Extract the attached zip. cd into the extracted folder and run the synthesis script yosys -c synth.tcl.

Expected Behavior

Clean exit

Actual Behavior

Looks like a crash

[Ravenslofty]

First things first: running the testcase results in "yosys-abc: src/map/scl/sclLibUtil.c:77: void abc::Abc_SclHashCells(SC_Lib*): Assertion '*pPlace == -1' failed." because you're passing -liberty $sky130_lib twice in the arguments to abc; I'm pretty sure you didn't mean to do that.

Also, though it's harmless, your script includes e.g. retime,-D,{D}, when {D} expands to include -D N already, resulting in abc getting -D -D N.

The actually reported issue is instead "ERROR: ABC: execution of command ""/usr/local/bin/yosys-abc" -s -f /tmp/yosys-abc-TaAwGJ/abc.script 2>&1" failed: return code 139." after stime -p.

After writing a wrapper script for abc that calls gdb, I get this:

Program received signal SIGSEGV, Segmentation fault.
0x00005555559d42db in abc::Abc_SclTimeNtkPrint (p=p@entry=0x5555577b1410, fShowAll=fShowAll@entry=0,
    fPrintPath=fPrintPath@entry=1) at src/map/scl/sclSize.c:203
203                 nLength = Abc_MaxInt( nLength, strlen(Abc_SclObjCell(pObj)->pName) );
#0  0x00005555559d42db in abc::Abc_SclTimeNtkPrint (p=p@entry=0x5555577b1410, fShowAll=fShowAll@entry=0,
    fPrintPath=fPrintPath@entry=1) at src/map/scl/sclSize.c:203
#1  0x00005555559d70cc in abc::Abc_SclTimePerformInt (pLib=pLib@entry=0x55555686f9e0, pNtk=pNtk@entry=0x555557646440,
    nTreeCRatio=nTreeCRatio@entry=0, fUseWireLoads=fUseWireLoads@entry=0, fShowAll=fShowAll@entry=0,
    fPrintPath=fPrintPath@entry=1, fDumpStats=0) at src/map/scl/sclSize.c:690
#2  0x00005555559d73e0 in abc::Abc_SclTimePerform (pLib=0x55555686f9e0, pNtk=0x555557646440, nTreeCRatio=nTreeCRatio@entry=0,
    fUseWireLoads=fUseWireLoads@entry=0, fShowAll=fShowAll@entry=0, fPrintPath=fPrintPath@entry=1, fDumpStats=0)
    at src/map/scl/sclSize.c:712
#3  0x00005555559a7472 in abc::Scl_CommandStime (pAbc=0x555556838200, argc=2, argv=0x5555578a8d20) at src/map/scl/scl.c:895
#4  0x00005555557a6c2b in abc::CmdCommandDispatch (pAbc=pAbc@entry=0x555556838200, pargc=pargc@entry=0x7ffffffed108,
    pargv=pargv@entry=0x7ffffffed110) at src/base/cmd/cmdUtils.c:157
#5  0x00005555557a078e in abc::Cmd_CommandExecute (pAbc=pAbc@entry=0x555556838200,
    sCommand=sCommand@entry=0x7ffffffed170 "stime -p;\n") at src/base/cmd/cmdApi.c:210
#6  0x000055555579e6d7 in abc::CmdCommandSource (pAbc=0x555556838200, argc=<optimized out>, argv=<optimized out>)
    at src/base/cmd/cmd.c:787
#7  0x00005555557a6c2b in abc::CmdCommandDispatch (pAbc=pAbc@entry=0x555556838200, pargc=pargc@entry=0x7fffffff5228,
    pargv=pargv@entry=0x7fffffff5230) at src/base/cmd/cmdUtils.c:157
#8  0x00005555557a078e in abc::Cmd_CommandExecute (pAbc=pAbc@entry=0x555556838200,
    sCommand=sCommand@entry=0x555556837e10 "source abc.script") at src/base/cmd/cmdApi.c:210
#9  0x00005555557e758a in abc::Abc_RealMain (argc=3, argv=<optimized out>) at src/base/main/mainReal.c:332
#10 0x00005555557e4e40 in main (argc=<optimized out>, argv=<optimized out>) at src/base/main/main.c:11

[Ravenslofty]

Reported upstream as berkeley-abc/abc#305.

[kareefardi]

@Ravenslofty Thanks for your help. Yes the double -liberty wasn't intentional.

Also, though it's harmless, your script includes e.g. retime,-D,{D}, when {D} expands to include -D N already, resulting in abc getting -D -D N.

Thanks for noticing that as well.