bash longjmp issue

Larry Hall (Cygwin) reply-to-list-only-lh@cygwin.com
Mon Dec 18 04:46:00 GMT 2006


<http://cygwin.com/acronyms/#TOFU>.  Reformatted.

K Kafka wrote:
> On 12/17/06, Larry Hall (Cygwin) <reply-to-list-only-lh <AT> cygwin <DOT> com> wrote:
>> K Kafka wrote:
>> > On 12/17/06, Larry Hall (Cygwin) <reply-to-list-only-lh <AT> cygwin <DOT> com> 
>> wrote:


<http://cygwin.com/acronyms/#PCYMTNQREAIYR>.  There's no need to feed the
spammers.  Thanks.


>> >> K Kafka wrote:
>> >> > Hi all,
>> >> >
>> >> > I have been looking around for a possible cause for my problem with
>> >> > bash. I left a "tail -f" running for a few days. When I came back,
>> >> > none of the service would run anymore (cygsrv, exim, sshd). After a
>> >> > bit of researching I've come accross quite a few posts but nothing
>> >> > concrete.
>> >> >
>> >> > It seems that bash.exe is crashing with a WFSO longjmp error 
>> after the
>> >> > 5min timeout. Preventing everything else. From some posts, it 
>> seems it
>> >> > could be coming from a device (webcam) or a software such as the
>> >> > anti-virus. I have almost nothing installed on this machine aside 
>> from
>> >> > MySQL and Tomcat5. Its a Xeon server running Windows 2003 server.
>> >> >
>> >> > Here is the actual error for bash.
>> >> > C:\cygwin\bin>      4 [main] bash 3376 C:\cygwin\bin\bash.exe: ***
>> >> > fatal error -  WFSO timed out after longjmp
>> >> >
>> >> > Some people have been suggesting to try a rebaseall. Same thing
>> >> > happens, I get two errors, a fhandler_base::fork_fixup 
>> immediately and
>> >> > an WFSO longjmp error after 5 min.
>> >> >
>> >> > Here are the errors:
>> >> > C:\cygwin\bin>      4 [main] ash 1328 fhandler_base::fork_fixup:
>> >> > /bin/rebaseall - Win32 error 5, handle io_handle<0x0>
>> >> >
>> >> > C:\cygwin\bin>      4 [main] ash 1856 C:\cygwin\bin\ash.exe: *** 
>> fatal
>> >> > error - WFSO timed out after longjmp
>> >> >
>> >> > I've also updated cygwin to the latest version and reinstalled bash
>> >> > and cygsrv. No changes, same errors.
>> >> >
>> >> > If someone has an idea, please let me know. I'm at a loss and can't
>> >> > figure what to do next.
>> >>
>> >>
>> >> Let's start here:
>> >>
>> >> > Problem reports:       http://cygwin.com/problems.html
>> >>
>> >>
>> >> You've given us some information about what you see and when but we're
>> >> missing allot of specifics about your configuration and a possible 
>> test
>> >> case (or cases).  I'd recommend any follow-up posting to this list
>> >> contain the additional information outlined at the above web page.
>> >>
>>  > Hi again,
>>  >
>>  > Sorry about the lack of background info. Let's try to correct it. I've
>>  > attached the cygcheck output  and realized that there is a series of
>>  > "not found" that includes bash. Also, I was unfortunately not able to
>>  > reproduce a problem on another sytem.
>>  >
>>  > When double clicking on the cygwin icon (or launching bash --login -i
>>  > from the windows command line), the window hangs blank for 5 minutes
>>  > and then disappear. In the case of the windows command line, after a
>>  > few minutes it outputs the following error reported earlier:
>>  >
>>  > C:\cygwin\bin>bash.exe --login -i
>>  > C:\cygwin\bin>      4 [main] bash 3376 C:\cygwin\bin\bash.exe:
>>  > ***fatal error -  WFSO timed out after longjmp
>>  >
>>  > There is a few related records in the Windows Application Event 
>> Viewer:
>>  > - Faulting application id.exe, version 0.0.0.0, faulting module
>>  > cygwin1.dll, version 1005.22.0.0, fault address 0x0000365f.
>>  > - Faulting application cygrunsrv.exe, version 0.0.0.0, faulting module
>>  > cygwin1.dll, version 1005.22.0.0, fault address 0x0000365f.
>>  >
>>  > Also, when I try to start manually a service such as inetd or exim,
>>  > the error is:
>>  > - The CYGWIN inetd service terminated unexpectedly.  It has done 
>> this 2
>>  > time(s).
>>  > - The exim service terminated unexpectedly.  It has done this 2 
>> time(s).
>>  >
>>  > I hope this is a bit more precise. Thanks you.
>>
>>
>> Yep, that's good.  Looks like your installation didn't complete, probably
>> because you didn't stop your services before updating.  I'd suggest 
>> rebooting,
>> if you have not already since installing.  Otherwise, I'd recommend 
>> running
>> 'setup.exe' again and allow it to just run with the packages you have
>> already selected.  This will run the postinstall scripts, which seem
>> to have not run the last time.  If that still doesn't work, look in
>> '/etc/postinstall' and look for scripts that don't have the suffix 
>> '.done'.
>> You'll want to run each of these manually with 'c:\cygwin\bin\bash.exe'.
>> That's what rerunning 'setup.exe' should do for you but doing so manually
>> should give you more info about any problems.  You can also look in
>> '/var/log/setup.log.full' for diagnostic messages that were generated by
>> 'setup.exe' the last time you ran it.  This should give you some 
>> information
>> about what postinstall scripts had complaints.
>>
> Hi Larry,
> 
> Thank you. So you are right, I did not stop any of the services before
> upgrading. I'll know it in the future.
> 
> On the actual problem:
> 1- I've rebooted since the last install, the problem persists
> 2- All the files in /etc/postinstall have the .done suffix
> 3- In setup.log.full, the only error is about McShield (Could not open
> service McShield for query, start and stop). Is this related?


No.  That just means you don't have McAfee.


> So, I've followed your suggestion and
> 1- relaunched setup.exe, nothing was installed (mbox note: Nothing
> needed to be installed)
> 2- I've rebooted once more nothing changed (i.e. same errors with bash)
> 3- Actually, when I logged again on the box there was a few errors:
>       3x:  id.exe encountered a problem and needed to close:
> szAppName : id.exe     szAppVer : 0.0.0.0     szModName : cygwin1.dll
> szModVer : 1005.22.0.0     offset : 0000365f
>       3x:  cygrunsrv.exe encountered a problem and needed to close:
> szAppName : cygrunsrv.exe     szAppVer : 0.0.0.0     szModName :
> cygwin1.dll
> szModVer : 1005.22.0.0     offset : 0000365f
> 3- when checking the cygcheck output once more I realized that
> something was written to stderr: "garbled output from 'id' command -
> no uid= found". Can this be related?


Kind of.  It's the result of the failure of id.exe above.


> Then, tryed to reinstall bash, cygwin, cygutils and cygrunsrv using 
> setup.exe
> 
> 1- in setup.log.full there is a few errors:
> 
>  2006/12/17 18:43:06 running: C:\cygwin\bin\bash.exe -c
> /etc/postinstall/00bash.sh
>  2006/12/17 18:43:06 error: Unable to remove temporary file
> '/var/log/setup.log.postinstallXa03412'
>  2006/12/17 18:43:06 running: cmd.exe /c 
> C:\cygwin\etc\postinstall\01bash.bat
>  2006/12/17 18:43:07 abnormal exit: exit code=0
>  2006/12/17 18:43:07 error: Unable to remove temporary file
> '/var/log/setup.log.postinstallXa03412'
> 
> The 00bash.sh and 01bash.bat don't exist anymore so I can't run them
> manually. The Xa03412 file was deleted manually afterward just in
> case.
> 
> 2- Bash gives the same errors, /etc/postinstall files have all the .done 
> suffix
> 
> Should I try to uninstall completely and reinstall?

You could try it.  Shouldn't be necessary though.  Still it might be the
easiest way to a solution.  Otherwise, you can try running 01bash.bat from
cmd.exe making sure echo is on and watch for problems relative to the
statements being executed.

-- 
Larry Hall                              http://www.rfk.com
RFK Partners, Inc.                      (508) 893-9779 - RFK Office
216 Dalton Rd.                          (508) 893-9889 - FAX
Holliston, MA 01746

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/



More information about the Cygwin mailing list