Description of problem: test abort01 if hang with 2.4.21-15 kernel at amd Athlon. Version-Release number of selected component (if applicable): kernel-2.4.21-15.EL kernel-unsupported-2.4.21-15.EL glibc-devel-2.3.2-95.20 glibc-utils-2.3.2-95.20 glibc-kernheaders-2.4-8.34 glibc-common-2.3.2-95.20 glibc-debug-2.3.2-95.20 glibc-2.3.2-95.20 glibc-headers-2.3.2-95.20 glibc-profile-2.3.2-95.20 How reproducible: install LTP test suite and run test. Steps to Reproduce: 1. 2. 3. Actual results: test abort01 is hang. Expected results: test suite work ok. Additional info: <<<test_start>>> tag=abort01 stime=1084870400 cmdline="ulimit -c 1024;abort01" contacts="" analysis=exit initiation_status="ok" <<<test_output>>> .................... 16075 pts/8 S 0:00 /bin/sh ./runalltests.sh 16175 pts/8 S 0:00 /root/temp/2/ltp/pan/pan -e -S -a 16075 -n 16075 -f / 16176 pts/8 T 0:00 sh -c ulimit -c 1024;abort01 16177 pts/8 T 0:00 abort01
Please update with the following information: 1 - Version of LTP that you are using to produce the abort01 test hang condition. 2 - Output from "uname -a".
1. LPT is last cvs version, updated yesterday. 2. [root@berloga root]# uname -a Linux berloga.shadowland 2.4.21-15.EL #1 Wed May 12 16:17:59 EEST 2004 i686 athlon i386 GNU/Linux PS. hang is in execute any tests group in batch. if run test manualy test finished ok.
The manual test ran fine on our Athlon test system last night. Your hang condition is only when LTP is run in batch, e.g. batch -f ./runalltests.sh, correct?
yes. but one - i don`t use batch -f. i direct start runalltests.sh or start via ltpmenu.
Tell me what you mean by direct start please. You are at the bash prompt and you enter "./runalltests.sh"? You run the command as root? If this is the case then I do not see the hang condition on my Athlon based system. I can put the LTP software package I used on a http site. Would you be interested in using that software to verify your hang condition?
1. yes - i login as root - start xfree with kde desktop and in kconsole start ./runalltest.sh 2. ok. put you LTP package at web and i try to run it at my system. if it not been hanged - it LTP bug. or i can put my LTP directory for you download.
Please use the LTP software located at http://people.redhat.com/peterm/ltp/. Please update bug entry with the results of your test.
with that lpt distributions tests worked as expected. Thanks for you work. Сan I get from you news versions of LTP adapted for RH ?
If the April version meets your requirements then please use it for the short term. The next step is for me to use the latest unmodified version from LTP to see if it hangs on my Athlon system. Update with download information so I can get a copy of your exact LPT version that is hanging. I will follow up with my findings.
I try to use last LTP version available at LTP site. http://prdownloads.sourceforge.net/ltp/ltp-full-20040506.tgz?download or you can download my cvs snapshot http://st50.sevcity.net/umka/ltp-orig.tar.gz All of it`s copy's hanged at start tests.
Created attachment 100410 [details] task trace log
It`s look a strange but today my workstation hanged again. I tried start you LTP suite and test locked again. I enable sysrq and record to syslog tasks trace.
What are progress with this bug? You have any patches which i can test? I found one bug in signal transmit and it fixed in 2.6 (bug in __kill_pg_info) but can`t see it fix in RH kernel. Why? Also I found Roland McGrath <roland> posted to linux-kernel@ mail list "[PATCH] bogus sigaltstack calls by rt_sigreturn" who affected to RH kernel, but RH distribution not include it. Why?
I updated my system to RH EL3 update3 via up2date. kernel 2.4.21-20 and glibc # rpm -qa | grep glibc glibc-kernheaders-2.4-8.34.1 glibc-common-2.3.2-95.27 glibc-2.3.2-95.27 glibc-profile-2.3.2-95.27 glibc-devel-2.3.2-95.27 glibc-headers-2.3.2-95.27 glibc-utils-2.3.2-95.27 but bug still here. What information need for fix it ?
hm.. know i try use ltp again, and found what is trigered this bug. If i run ltp tests from bash shell it run ok, but if i start runtest.sh from mc 4.6.0 i seen this bug.
*** Bug 146023 has been marked as a duplicate of this bug. ***
This bug exist at last RH EL4 kernel - 2.6.9-11.
Closing out this request as notabug for RHEL3, ltp from bash works properly.