Well, I switched to Dapper Drake Flight 6, but the X errors persist. I can reproduce them trivially tonight on my laptop by hitting Find in wine's Notepad (I was trying to reproduce http://bugs.winehq.org/show_bug.cgi?id=1614). The errors look like
X Error of failed request: BadAtom (invalid Atom parameter) Major opcode of failed request: 17 (X_GetAtomName) Atom id in failed request: 0x0
and now that winedbg is working again (yay), setting WINEDEBUG=+synchronous generates a good backtrace: 3 0x7f231cea _XError+0xf4 in libx11.so.6 (0x7f231cea) 4 0x7f23229e _XReply+0x185 in libx11.so.6 (0x7f23229e) 5 0x7f212c7d XGetAtomNames+0x131 in libx11.so.6 (0x7f212c7d) 6 0x7f315cfd X11DRV_CLIPBOARD_InsertSelectionProperties+0x17b(properties=<register not in topmost frame>) [/home/dank/wine/dlls/x11drv/clipboard.c:1788] in winex11 (0x7f315cfd) 7 0x7f3182ae X11DRV_CLIPBOARD_UpdateCache+0x3e6(lpcbinfo=<register not in topmost frame>) [/home/dank/wine/dlls/x11drv/clipboard.c:1911] in winex11 (0x7f3182ae) 8 0x7f3187a8 X11DRV_IsClipboardFormatAvailable+0x28(wFormat=0x1) [/home/dank/wine/dlls/x11drv/clipboard.c:2498] in winex11 (0x7f3187a8) 9 0x7f6c30f9 IsClipboardFormatAvailable+0x25(wFormat=0x1) [/home/dank/wine/dlls/user/clipboard.c:583] in user32 (0x7f6c30f9) 10 0x7fb84488 NOTEPAD_WndProc+0x2e8(hWnd=0x70124, msg=0x117, wParam=0x1e0, lParam=0x2) [/home/dank/wine/programs/notepad/main.c:160] in notepad (0x7fb84488)
Turning on +clipboard makes the error go away, so I guess it's a Heisenbug. (Doesn't make find work, though; that's http://bugs.winehq.org/show_bug.cgi?id=1614 )
Running notepad.exe.so under valgrind reveals a few suspicious lines:
==22697== Conditional jump or move depends on uninitialised value(s) ==22697== at 0x205EE469: SELECTOR_SetEntries (../../include/wine/library.h:130) ==22697== ==22697== Conditional jump or move depends on uninitialised value(s) ==22697== at 0x20604A9C: WOWTHUNK_Init (../../include/wine/library.h:130) ==22697== ==22697== Conditional jump or move depends on uninitialised value(s) ==22697== at 0x219EFD36: XcursorImageHash (in /usr/lib/libXcursor.so.1.0.2) ==22697== ==22697== Syscall param write(buf) points to uninitialised byte(s) ==22697== at 0x20000772: (within /lib/ld-2.3.6.so) ==22697== Address 0x213A4EF2 is on thread 1's stack ==22697== ==22697== Conditional jump or move depends on uninitialised value(s) ==22697== at 0x200235AD: wine_ldt_get_entry (../../include/wine/library.h:130) ==22697== ==22697== Conditional jump or move depends on uninitialised value(s) ==22697== at 0x2180CC68: X11DRV_KeymapNotify (keyboard.c:1275) ==22697== ==22697== Conditional jump or move depends on uninitialised value(s) ==22697== at 0x2180CC9E: X11DRV_KeymapNotify (keyboard.c:1278)
None of the above happen after the ^F, though, so while they look bad, they might not have anything to do with notepad's problems.
Time to sleep. I'm posting these notes for my own use later, and in case anyone feels like checking up on any of the BadAtom, valgrind, or 'find in notepad' errors. - Dan
-- Wine for Windows ISVs: http://kegel.com/wine/isv
On 4/7/06, Dan Kegel dank@kegel.com wrote:
Well, I switched to Dapper Drake Flight 6, but the X errors persist. I can reproduce them trivially tonight on my laptop by hitting Find in wine's Notepad
Simpler way to reproduce: run the clipboard test.
~/wine/dlls/user/tests$ ~/wine/wine user32_test.exe.so clipboard clipboard.c:123: Test failed: GlobalGetAtomNameA should fail clipboard.c:124: Test failed: Last error should be set to 6, not -559038737 X Error of failed request: BadAtom (invalid Atom parameter) Major opcode of failed request: 17 (X_GetAtomName) Atom id in failed request: 0x0 Serial number of failed request: 40 Current serial number in output stream: 42
Again, this is with both Ubuntu 05.10 and Dapper Drake flight 6; on Dapper, the x server is xserver-xorg version 7.0.0-0ubuntu25. - Dan
- Dan
On 4/8/06, Dan Kegel dank@kegel.com wrote:
Well, I switched to Dapper Drake Flight 6, but the X errors persist. I can reproduce them trivially tonight on my laptop by hitting Find in wine's Notepad
Simpler way to reproduce: run the clipboard test.
I think my BadAtom errors are an instance of http://bugs.winehq.org/show_bug.cgi?id=4601 I attached a cheesy patch to that bug report; the patch just ignores the bad atoms (which are all zero). It'd be better to see why the bad atoms are getting sent, but maybe this patch will let me keep testing without Firefox crashing hard all the time for me.
-
Dan Kegel