Hi Roland, Roland Mainz p??e v po 02. 03. 2009 v 18:50 +0100: > Milan Jurik wrote: > > V po, 02. 03. 2009 v 17:25, Roland Mainz p????e: > > > Milan Jurik wrote: > > > > V po, 02. 03. 2009 v 16:38, Alan Coopersmith p?f??????e: > > > > > <http://defect.opensolaris.org/bz/show_bug.cgi?id=7009> > > > > > > > > thank you. I can reproduce it on build 108. > > > > > > How did you reproduce the problem ? I've tried to reproduce the crash on > > > SXCE B106 without sucesss... ;-( > > > ... or is OpenSolaris/Indiana a requirement ? > > > > Step-by-step on Indiana. I will try later at home on my personal SXCE > > b109. >
And even my SXCE b109 has the same result... The same isalist I sent to you. > I tried this on my B106/x86 machine and get no crash: > -- snip -- > print -n 2 > print -n2 > print -u 2 Test > print -u 2 Test >/dev/null > print -u 2 Test > tee /dev/null > print -u 2 Test > tee -a /dev/null > print -u 2 "Test" > tee /tmp/test > tee /tmp/test; print "Test" > rm tee > print -u 2 "Test" | tee /tmp/test > true > -- snip -- > > > > > I can create Bugster CR, if > > > > there is no automagic for it available. > > > > > > Please create one and mark the bug at http://defect.opensolaris.org as > > > TRACKEDINBUGSTER > > > > Done. You can download corefile from defect.o.o > > Thanks! :-) > > Stack trace from attachment.cgi?id=1447 looks like this: > -- snip -- > core 'xxx' of 9456: /bin/ksh > fffffd7ffede0173 sfdisc () + 51b > fffffd7ffeea23fb b_tee () + 263 > fffffd7ffef5ea23 sh_exec () + 2deb > fffffd7ffef5f78c sh_exec () + 3b54 > fffffd7ffef5d9c8 sh_exec () + 1d90 > fffffd7ffef36e86 exfile () + 786 > fffffd7ffef36676 sh_main () + 7fe > 0000000000400e72 main () + 52 > 0000000000400ccc ???????? () > -- snip -- > > Stack trace from attachment.cgi?id=1448 looks like this: > -- snip -- > core 'xxx' of 9476: /bin/ksh > fffffd7ffede725a sfraise () + 172 > fffffd7ffedddc5f sfclose () + 187 > fffffd7ffef204de sh_close () + 46 > fffffd7ffef203a7 sh_iorenumber () + bf > fffffd7ffef5f3c4 sh_exec () + 378c > fffffd7ffef5d943 sh_exec () + 1d0b > fffffd7ffef36e86 exfile () + 786 > fffffd7ffef36676 sh_main () + 7fe > 0000000000400e72 main () + 52 > 0000000000400ccc ???????? () > -- snip -- > > I need a testcase to reproduce the issue to get more > > Milan: Do you get the same stacktraces for the 32bit binary, e.g. > /usr/bin/i86/ksh93 ? And here it is in forced (/i86/ subdir) 32-bit mode: > $c 0x8067635(fecef068, 4, 0, fec75fde) libast.so.1`sfclose+0x195(fecef068, fef83540, 20a, fedad76e) libshell.so.1`sh_close+0x46(1, 20a, 804611c, fedad5a6) libshell.so.1`sh_iorenumber+0xda(fee16d68) libshell.so.1`sh_exec+0x3a51(80675ec, 4, 806cb60, fede4255) libshell.so.1`sh_exec+0x1f2f(8067674, 4, 0, fedc1b45) libshell.so.1`exfile+0x716(fee16d68, fecef0d0) libshell.so.1`sh_main+0x808(1, 8046f98, 0, fee60ed2, 4, 0) main+0x4a(1, 8046f98, 8046fa0, 8046f8c) _start+0x7d(1, 80470e8, 0, 8061218, 8061248, 8047230) Btw. I would guess (based on 2 ksh93 core files created in the same second with only one core dump message) that there are two bugs, one the result of this sequence of commands and the second as result of coredump of forked ksh93. But I need to investigate that idea a bit more yet. And there is really something wrong, I tried to minimize the test case, see this log: mj162486 at xylabone:~$ ksh93 mj162486 at xylabone:~$ print -u 2 "Test" > tee /tmp/test Test /tmp/test mj162486 at xylabone:~$ cat /tmp/test mj162486 at xylabone:~$ tee /tmp/test; print "Test" - here I used Ctrl+c mj162486 at xylabone:~$ rm tee mj162486 at xylabone:~$ print -u 2 "Test" | tee /tmp/test Segmentation Fault (core dumped) mj162486 at xylabone:~$ mj162486 at xylabone:~$ mj162486 at xylabone:~$ ksh93 mj162486 at xylabone:~$ tee /tmp/test; print "Test" mj162486 at xylabone:~$ mj162486 at xylabone:~$ rm /tmp/test mj162486 at xylabone:~$ ksh93 mj162486 at xylabone:~$ touch /tmp/test mj162486 at xylabone:~$ tee /tmp/test; print "Test" mj162486 at xylabone:~$ rm tee rm: tee: No such file or directory mj162486 at xylabone:~$ - here I put Ctrl+d Segmentation Fault (core dumped) > $c 0x41c0b0() libast.so.1`sfsetbuf+0x3f8() libast.so.1`_sfcleanup+0x16c() libc.so.1`_exithandle+0x61() libc.so.1`exit+0x11() libshell.so.1`sh_done+0x1c4() libshell.so.1`sh_main+0x80c() main+0x52() 0x400ccc() So another bug, I will fill it later. Some problems in cleanup during exit. All is from SXCE 109. Best regards, Milan
