Bugzilla – Full Text Bug Listing
| Summary: | VM Crash when running remotely via cygwin/X | ||||||
|---|---|---|---|---|---|---|---|
| Product: | [Eclipse Project] Platform | Reporter: | David Martinez <hackerdude.david> | ||||
| Component: | SWT | Assignee: | Billy Biggs <billy.biggs> | ||||
| Status: | RESOLVED FIXED | QA Contact: | |||||
| Severity: | normal | ||||||
| Priority: | P3 | CC: | billy.biggs | ||||
| Version: | 3.0.1 | ||||||
| Target Milestone: | 3.1 M6 | ||||||
| Hardware: | Sun | ||||||
| OS: | Solaris-Motif | ||||||
| Whiteboard: | |||||||
| Attachments: |
|
||||||
Does your version of ssh have a -Y option for forwarding with different authentication? I have heard of some versions of ssh being too picky about security, and breaking applications. Turning on the -Y version does the following: 1) Show a "Warning: No xauth data; using fake authentication data for X11 forwarding" message upon successful ssh negotiation. 2) Provide me with the select workbench screen 3) Fail to respond to events (keyboard, mouse, etc). Resizing also seems to not respond in the way resizing in X manages to not respond :-) (i.e. you can resize the window, but the bounds recalculation never takes place). Close events are also ignored, making you press Control-C on the console as the only way to exit. Both Netscape and SwingSet demo seem to work fine under both -X and -X -Y. To be clear, you get further when using -Y than with -X? I heard this before but was never clear on what exactly is being blocked with -X that is not with -Y. Bug 82020 describes some problems with the cygwin window manager and eclipse. Is this similar to what you are seeing? Created attachment 18217 [details]
Modified simple Hello program for testing
Test I am using to limit to SWT.
I copied the test from Bug 82020 and tried it on my configuration. What I get is that the screen shows up ok, no infinite loop (The "Iteration" loop never prints anything). But there are no events being dispatched that I can see. It just locks up tight on shell.open(). If I change shell.open and put: System.out.println("About to make visible " ); shell.setVisible(true); System.out.println("now it's visible " ); Then I only get the first println. Same problem as Comment #6 in that bug. https://bugs.eclipse.org/bugs/show_bug.cgi?id=82020#c6 Sorry, never answered to your question above: To be clear, you get further when using -Y than with -X? Yes. :-) The memmove problem goes away with -Y, and now I'm presented with the workspace selection dialog. But it doesn't do anything. I would like to keep these two bugs separate. If we're crashing with -Y and not with -X, then this is interesting. I am very curious to know what is causing this crash. Do you know what the change in policy is between -Y and -X? The last time I looked this up, there was no documentation describing what X calls might be restricted. The other bug with the blocking on open() is that we're waiting for the window to be mapped, and not getting any map event (even though the window is clearly mapped). This sounds like a window manager problem, although maybe one we should be protecting against. Alexander Gottwald is a developer on the Xming project and tells me he will look at the problem, but has been busy with school lately. Please add a comment and your test case for this issue to bug 82020. I went digging through for this, and apparently the Image init constructor is obtaining an untrusted .Xauthority cookie. The -X option by default no longer assumes that the cookies of the children of the SSH connection are trusted, so an untrusted cookie is generated. Could it be that the Solaris image class is trying to get a trusted cookie for the image and receiving null (hence memmove issue)? Here is what prompted the -Y flag and default cookie handling change: http://www.securityfocus.com/bid/1006/discussion/ The new SSH assumes all clients are untrusted, and denies access to the X client. So basically, the -X option now means that all cookies you ask for are untrusted (i.e, don't give you access to the client machine), while -Y allows for the cookies to be trusted. I know very little about X. Is there a programmatic way of explicitly requesting a cookie to be trusted? I am also unable to find a discussion on the specific policies and permissions anywhere, but from the likes of this image display must be about manipulating data that "belongs to the client". from the xauth (1) man page (cygwin) "..If the trusted option is used, clients that connect using this authorization will have full run of the display, as usual. If untrusted is used, clients that connect using this authorization will be considered untrusted and prevented from stealing or tampering with data belonging to trusted clients. See the SECURITY extension specification for full details on the restrictions imposed on untrusted clients. The default is untrusted." The question is, does the image display belong to the client or the server? As for the other bug, I'll go ahead and add my testcase/comment. Thanks! It's not really like that, X applications do not deal directly with cookies. I did some research into what ssh does and what caused this crash. ssh -X marks the X application as "untrusted", limiting the application's ability to access resources such as the root window, keyboard events, and focus events. The specifics of what is restricted is defined by your X server (in this case, cygwin), which makes it difficult to know exactly what will be disallowed. This crash is caused because in our Image creation code, we are using XGetImage on the root window (effectively a mini screenshot) to initialize a buffer at the same depth as the screen. This is being restricted by cygwin, we don't notice, and so we crash. By my reading of the SECURITY extension draft on the web, I can fix this, but I bet we will just crash somewhere else. It is too easy to write code that inadvertently relies on resources that are now restricted. I think if the ssh project intended to increase security by changing this behaviour, all they have really done is confused both users and developers by not providing information on how to fix applications, or what security is achieved by these options. For example: http://www.linuxquestions.org/questions/history/240073 http://hea-www.harvard.edu/RD/ds9/issue.html http://lists.freebsd.org/pipermail/freebsd-stable/2004-October/008979.html Crash should be fixed > 20050223 Please re-open if you still see this particular crash. If it crashes in another place when doing ssh -X, please open a new bug. |
Using Cygwin 1.5.12 (0.116/4/2) and forwarding X from a Solaris 3.0.1 eclipse, the system fails to start with the following error: ssh -X eclipse.sun.box bash-2.05$ ./eclipse An unexpected exception has been detected in native code outside the VM. Unexpected Signal : 11 occurred at PC=0xFF38058C Function=memcpy+0xB0 Library=/usr/platform/sun4u/lib/libc_psr.so.1 Current Java thread: at org.eclipse.swt.internal.motif.OS.memmove(Native Method) at org.eclipse.swt.graphics.Image.<init>(Image.java:245) at org.eclipse.swt.widgets.Label.setBitmap(Label.java:380) at org.eclipse.swt.widgets.Label.setImage(Label.java:462) at org.eclipse.jface.dialogs.TitleAreaDialog.createTitleArea(TitleAreaDialog.java:180) at org.eclipse.jface.dialogs.TitleAreaDialog.createContents(TitleAreaDialog.java:107) at org.eclipse.jface.window.Window.create(Window.java:348) at org.eclipse.jface.dialogs.Dialog.create(Dialog.java:925) at org.eclipse.jface.window.Window.open(Window.java:637) at org.eclipse.ui.internal.ide.ChooseWorkspaceDialog.prompt(ChooseWorkspaceDialog.java:65) at org.eclipse.ui.internal.ide.IDEApplication.promptForWorkspace(IDEApplication.java:314) at org.eclipse.ui.internal.ide.IDEApplication.checkInstanceLocation(IDEApplication.java:219) at org.eclipse.ui.internal.ide.IDEApplication.run(IDEApplication.java:83) at org.eclipse.core.internal.runtime.PlatformActivator$1.run(PlatformActivator.java:335) at org.eclipse.core.runtime.adaptor.EclipseStarter.run(EclipseStarter.java:273) at org.eclipse.core.runtime.adaptor.EclipseStarter.run(EclipseStarter.java:129) at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method) at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:39) at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:25) at java.lang.reflect.Method.invoke(Method.java:324) at org.eclipse.core.launcher.Main.basicRun(Main.java:185) at org.eclipse.core.launcher.Main.run(Main.java:704) at org.eclipse.core.launcher.Main.main(Main.java:688) Dynamic libraries: 0x10000 /usr/bin/java 0xff360000 /usr/lib/libthread.so.1 0xff39a000 /usr/lib/libdl.so.1 0xff280000 /usr/lib/libc.so.1 0xff380000 /usr/platform/SUNW,Ultra-80/lib/libc_psr.so.1 0xfec00000 /usr/j2se/jre/lib/sparc/client/libjvm.so 0xff240000 /usr/lib/libCrun.so.1 0xff210000 /usr/lib/libsocket.so.1 0xff100000 /usr/lib/libnsl.so.1 0xff1c0000 /usr/lib/libm.so.1 0xff0e0000 /usr/lib/libsched.so.1 0xff0c0000 /usr/lib/libmp.so.2 0xff060000 /usr/j2se/jre/lib/sparc/native_threads/libhpi.so 0xfebd0000 /usr/j2se/jre/lib/sparc/libverify.so 0xfeb90000 /usr/j2se/jre/lib/sparc/libjava.so 0xff030000 /usr/j2se/jre/lib/sparc/libzip.so 0xfe1d0000 /usr/j2se/jre/lib/sparc/libnet.so 0xfc6e0000 /usr/j2se/jre/lib/sparc/libnio.so 0xfc6c0000 /usr/lib/librt.so.1 0xfc6a0000 /usr/lib/libaio.so.1 0xfc5e0000 /usr/lib/libmd5.so.1 0xfc5c0000 /usr/lib/libsendfile.so.1 0xfbe90000 /export/home/davidm/eclipse/plugins/org.eclipse.swt.motif_3.0.1/os/solaris/sparc/libswt-motif-3063.so 0xf0d00000 /usr/lib/libXm.so.4 0xfbd90000 /usr/openwin/lib/libXt.so.4 0xf0c00000 /usr/openwin/lib/libX11.so.4 0xfbfc0000 /usr/openwin/lib/libXp.so.1 0xfbfa0000 /usr/openwin/lib/libXtst.so.1 0xfbcd0000 /usr/openwin/lib/libXext.so.0 0xfbcb0000 /usr/openwin/lib/libSM.so.6 0xfbc80000 /usr/openwin/lib/libICE.so.6 0xfbc60000 /usr/lib//liblayout.so Heap at VM Abort: Heap def new generation total 2112K, used 1429K [0xf1800000, 0xf1a20000, 0xf1f10000) eden space 2048K, 66% used [0xf1800000, 0xf19555f0, 0xf1a00000) from space 64K, 100% used [0xf1a10000, 0xf1a20000, 0xf1a20000) to space 64K, 0% used [0xf1a00000, 0xf1a00000, 0xf1a10000) tenured generation total 4680K, used 3521K [0xf1f10000, 0xf23a2000, 0xf5800000) the space 4680K, 75% used [0xf1f10000, 0xf22806f0, 0xf2280800, 0xf23a2000) compacting perm gen total 5120K, used 4994K [0xf5800000, 0xf5d00000, 0xf9800000) the space 5120K, 97% used [0xf5800000, 0xf5ce0b08, 0xf5ce0c00, 0xf5d00000) Local Time = Tue Feb 22 08:01:34 2005 Elapsed Time = 11 # # The exception above was detected in native code outside the VM # # Java VM: Java HotSpot(TM) Client VM (1.4.2_05-b04 mixed mode) # # An error report file has been saved as hs_err_pid5146.log. # Please refer to the file for further information. # Running a Linux X server against it works OK, however. If this is a cygwin/X problem, I would like some information on what specifically is going wrong here so I can report it to the X folks. It seems to be having trouble copying the memory for an image..?