bash failed to initialize on telnet/rsh/rlogin server
I'm trying to get inetd set up for telnet/rsh/rlogin on my machine and have everything going up to the point that after login.exe gives the motd to the telnet client, a dialog window pops up on the server titled "bash.exe - Application Error" and says "The application failed to initialize properly (0xc022)." Acknowledging the dialog closes the connection with the client. I have telnetd working on another machine (NT4) but haven't been updating it to the latest cygwin packages. The new machine is running W2K, was updated this afternoon, and although there are 56 differences in the cygcheck -c outputs, these are the ones that seem relevant to me: (differences highlighted with *) Cygwin Package Information Cygwin Package Information Package Version Package Version * bash2.05a-3 bash2.05a-2 * cygwin 1.3.10-1cygwin 1.3.6-6 inetutils 1.3.2-17inetutils 1.3.2-17 login 1.4-3 login 1.4-3 Both are running with ntsec, and have identical passwd and group files (both from the domain server, passwd trimmed to a couple users plus the usual system accounts) Even though I thought I recalled seeing something about this in the last couple months, I've searched the mailing list for all kinds of combinations of keywords, including 'bash "failed to initialize"', and got no relevant hits. Before I try to debug this on my own, I wanted to check to see if there was some obvious answer (aside from "cygwin 1.3.10-1 broke this...because we're mean" :-) ) that I'm overlooking. And if I do need to debug this, where would I put the strace? in inetd.conf? if so, how? Thanks, Chris -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
bash failed to initialize on telnet/rsh/rlogin server
On Tuesday 26 Mar 02, Polley Christopher W writes: > I'm trying to get inetd set up for telnet/rsh/rlogin on my machine and have > everything going up to the point that after login.exe gives the motd to the > telnet client, a dialog window pops up on the server titled "bash.exe - > Application Error" and says "The application failed to initialize properly > (0xc022)." Acknowledging the dialog closes the connection with the > client. > > I have telnetd working on another machine (NT4) but haven't been updating it > to the latest cygwin packages. The new machine is running W2K, was updated > this afternoon, and although there are 56 differences in the cygcheck -c > outputs, these are the ones that seem relevant to me: (differences > highlighted with *) > > > Cygwin Package Information Cygwin Package Information > Package Version Package Version > * bash2.05a-3 bash2.05a-2 > * cygwin 1.3.10-1cygwin 1.3.6-6 > inetutils 1.3.2-17inetutils 1.3.2-17 > login 1.4-3 login 1.4-3 > > Both are running with ntsec, and have identical passwd and group files (both > from the domain server, passwd trimmed to a couple users plus the usual > system accounts) This happened to me on NT4 when I upgraded to cygwin-1.3.10, and was *not* using ntsec. Using ntsec fixed it. Is ntsec set early enough for inetd? (And are you rebooting when making changes?) Does the inetd service have sufficient user rights? Is everything readable/executable that needs to be? Double check your passwd/group entries with latest mkpasswd & mkgroup, possibly even from the latest inst snapshot. Make sure the user's groups are all in /etc/group. These have all been suggested in the past as possibly causing this error. Why I get the error with nontsec and not with ntsec, is still mystery, but Corinna has been on holiday. :-) Regards, David -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
RE: bash failed to initialize on telnet/rsh/rlogin server
Thanks, David, >This happened to me on NT4 when I upgraded to cygwin-1.3.10, and was >*not* using ntsec. Using ntsec fixed it. Is ntsec set early enough >for inetd? (And are you rebooting when making changes?) Does the >inetd service have sufficient user rights? Is everything I had ntsec set... but had not rebooted since setting it. Rebooted, and now all is well :-) (closing all cygwin processes including 'net stop inetd' wasn't enough) So now that I know how to isolate the problem, is there a way to strace a daemon? Warm regards, Chris -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
Re: bash failed to initialize on telnet/rsh/rlogin server
On Wed, Mar 27, 2002 at 09:05:20AM -0600, Polley Christopher W wrote: >So now that I know how to isolate the problem, is there a way to strace >a daemon? strace -p pid cgf -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
Re: strace on inetd (was RE: bash failed to initialize on telnet/rsh/ rlogin server)
On Wed, Mar 27, 2002 at 02:20:35PM -0600, Polley Christopher W wrote: >Are there more complete instructions to strace besides --help, the Cygwin >user guide, or winsup/utils/utils.sgml (all essentially the same thing)? If we had better documentation why would it be hidden? The best documentation is the source code. In this case, the use of strace -p would require that you do your debugging from an account which had the ability to access the executing pid. If you are running inetd from a privileged account and stracing from a user account that won't work for hopefully obvious reasons. Also, due to limitations in the Windows API, there is no way to strace a forked process via strace -p. strace -p uses the DebugActiveProcess function and there is no way to tell the process being debugged that you need to follow CreateProcess calls. cgf -- Please do not send me personal email with cygwin questions. Use the resources at http://cygwin.com/ . -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
Re: strace on inetd (was RE: bash failed to initialize on telnet/rsh/ rlogin server)
"Christopher Faylor" <[EMAIL PROTECTED]> wrote in message [EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > On Wed, Mar 27, 2002 at 02:20:35PM -0600, Polley Christopher W wrote: > >Are there more complete instructions to strace besides --help, the Cygwin > >user guide, or winsup/utils/utils.sgml (all essentially the same thing)? > > If we had better documentation why would it be hidden? I'm sure Cygwin tries to include the best possible documentation. But if someone can't find an answer in the official docs, it seems to me that asking for other sources is the right thing to do. For any number of reasons, there may be other sources of information that are not yet included in the official Cygwin documentation. > The best documentation is the source code. That depends on one's definition of "best". Source is certainly the most accurate documentation, but it is probably not the easiest to use. I'm sure the Cygwin developers are very busy and don't have a lot of time to offer free end-user support, especially since Cygwin does offer paid support options. And I do see a number of helpful articles posted by Cygwin developers. But occasionally I'm surprised at how hostile some answers are to what seem to be reasonable questions and requests. -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
RE: strace on inetd (was RE: bash failed to initialize on telnet/rsh/ rlogin server)
> -Original Message- > From: Matt Seitz [mailto:[EMAIL PROTECTED]] > Sent: Thursday, March 28, 2002 12:32 PM > > If we had better documentation why would it be hidden? > > I'm sure Cygwin tries to include the best possible > documentation. But if someone can't find an answer in the > official docs, it seems to me that asking for other sources > is the right thing to do. For any number of reasons, there > may be other sources of information that are not yet included > in the official Cygwin documentation. If Cygwin was a corporate project, then that would be true. However, it's not -IMO- a corporate project. Few, if any resources are allotted to Cygwin by RedHat. Of the folk who commit code to cygwin, only Corinna and Chris are RedHat employees, and AFAIK neither are paid-to-develop-cygwin. For a community project, there are few-if-any reasons for doco to be pending. It usually gets posted immediately, as that is the only way we get recognition. > > The best documentation is the source code. > > That depends on one's definition of "best". Source is > certainly the most accurate documentation, but it is probably > not the easiest to use. That's a very valid comment. However, IIRC you didn't define the metric of 'betterness' that you meant, so I think that Chris assuming better==accurate is quite reasonable. > I'm sure the Cygwin developers are very busy and don't have a > lot of time to offer free end-user support, especially since > Cygwin does offer paid support options. As I mention above... what Cygwin developers :}. Seriously though, while busy with other things, I usually still spend time on support, offering insights to new contributors, offering sarcasm to newbies , contributing bits of code... > And I do see a > number of helpful articles posted by Cygwin developers. But > occasionally I'm surprised at how hostile some answers are to > what seem to be reasonable questions and requests. I can't really comment on the other developers, but I find some behaviours (net shown by you AFAICT) very frustrating. * evidence of a lack of self-start-capability. (i.e. where do I find x, for some x documented on the web pages or high up on google searches) * 'it would be nice'. Particularly 'why doesn't x get improved'. * 'cygwin is broken'... when it passes the ltp test suite for that feature. Rob -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
Re: strace on inetd (was RE: bash failed to initialize on telnet/rsh/ rlogin server)
On Wed, 27 Mar 2002 19:20:19 -0500, you wrote: >On Wed, Mar 27, 2002@02:20:35PM -0600, Polley Christopher W wrote: >>Are there more complete instructions to strace besides --help, the Cygwin >>user guide, or winsup/utils/utils.sgml (all essentially the same thing)? > >If we had better documentation why would it be hidden? I didn't think it was "hidden", I only humbly thought there might be a more detailed document that the more clueful users on this list had awareness of. I'm sorry if I've offended you by my query. > >The best documentation is the source code. In this case, the use of >strace -p would require that you do your debugging from an account which >had the ability to access the executing pid. If you are running inetd >from a privileged account and stracing from a user account that won't >work for hopefully obvious reasons. Well, inetd is running from the system account (18). The user account I was using to run strace has local (machine) administrator privileges (i.e. is in the machine's "Administrators" group.) Is this sufficiently privileged? Is the silent failure (immediate exit from strace) I observed a result of this lack of privilege? Could I instead put telnet stream tcp nowait root /usr/bin/strace strace -o /strace_telnetd.log all /usr/bin/in.telnetd in inetd.conf? Or will this cause problems? > >Also, due to limitations in the Windows API, there is no way to strace a >forked process via strace -p. strace -p uses the DebugActiveProcess >function and there is no way to tell the process being debugged that you >need to follow CreateProcess calls. I'm interpreting your statement to mean that the -p and -f options of strace are incompatible. I see now from reviewing the code how this is so. I am confused by the lack of "couldn't attach to pid %d<%d> for debugging" that I would expect if DebugActiveProcess were to fail due to lack of permission, and the lack of "Attached to pid %d (windows pid %u)\n" that I would expect if it were to succeed (in the first 5 cases of my previous email) and the lack of strace output from inetd handling a telnet request (in the last case) (granted, it wouldn't output anything from any of inetd's children). I guess I'll need to attach gdb to strace attaching to a pid... Thanks, Chris -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
Re: strace on inetd (was RE: bash failed to initialize on telnet/rsh/ rlogin server)
At 02:09 AM 3/28/2002, Chris Polley wrote: >Well, inetd is running from the system account (18). The user account >I was using to run strace has local (machine) administrator privileges >(i.e. is in the machine's "Administrators" group.) Is this >sufficiently privileged? No. Look at inetutils-1.3.2.README. Larry Hall [EMAIL PROTECTED] RFK Partners, Inc. http://www.rfk.com 838 Washington Street (508) 893-9779 - RFK Office Holliston, MA 01746 (508) 893-9889 - FAX -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
Re: strace on inetd (was RE: bash failed to initialize on telnet/rsh/ rlogin server)
On Thu, Mar 28, 2002 at 01:09:29AM -0600, Chris Polley wrote: >On Wed, 27 Mar 2002 19:20:19 -0500, you wrote: >>On Wed, Mar 27, 2002@02:20:35PM -0600, Polley Christopher W wrote: >>>Are there more complete instructions to strace besides --help, the >>>Cygwin user guide, or winsup/utils/utils.sgml (all essentially the same >>>thing)? >> >>If we had better documentation why would it be hidden? > >I didn't think it was "hidden", I only humbly thought there might be a >more detailed document that the more clueful users on this list had >awareness of. I'm sorry if I've offended you by my query. Give me a break. I was making a simple statement. >>The best documentation is the source code. In this case, the use of >>strace -p would require that you do your debugging from an account which >>had the ability to access the executing pid. If you are running inetd >>from a privileged account and stracing from a user account that won't >>work for hopefully obvious reasons. >Well, inetd is running from the system account (18). The user account >I was using to run strace has local (machine) administrator privileges >(i.e. is in the machine's "Administrators" group.) Is this >sufficiently privileged? > >Is the silent failure (immediate exit from strace) I observed a result >of this lack of privilege? Don't know. >Could I instead put >telnet stream tcp nowait root /usr/bin/strace strace -o >/strace_telnetd.log all /usr/bin/in.telnetd >in inetd.conf? Or will this cause problems? Possibly. strace is not a cygwin program, though. You'd have to use MS-DOS paths and putting a non-cygwin program in there might cause different behavior. cgf -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/