Closed
Bug 36479
Opened 24 years ago
Closed 24 years ago
dual monitors under win98 crashes in GKGFXWIN.DLL
Categories
(Core :: XUL, defect, P3)
Tracking
()
VERIFIED
FIXED
People
(Reporter: grey, Assigned: mikepinkerton)
References
Details
(Keywords: crash, helpwanted, Whiteboard: [nsbeta2+] 6/1 (needs external verification))
From Bugzilla Helper: User-Agent: Mozilla/4.72 [en] (Win98; U) BuildID: M15 Release I have expereinced this crash in the M15 release build, and the nightly available during mid-day April 19. M14, M14 crypto, and NS6PR1 all worked fine. To isolate this problem, I have deleted all previous versions of Mozilla, along with the mozregistry.dat, and everything else I could fine everywhere, including rebooting after every crash to make sure it's not other system instabilities. I have checked the release notes, found no references to this, searched bugilla for anything related to GKGFXWIN and GKGFXWIN.DLL and came up empty. Both M15 and today's nightly do the exact same thing, and crash in the same area. Windows GPF data is as follows: MOZILLA caused an invalid page fault in module GKGFXWIN.DLL at 0177:60a41b34. Registers: EAX=00000000 CS=0177 EIP=60a41b34 EFLGS=00010246 EBX=0071fd40 SS=017f ESP=0068ef54 EBP=0068ef70 ECX=0068ef5c DS=017f ESI=00c6f660 FS=4de7 EDX=8000a728 ES=017f EDI=00000000 GS=0000 Bytes at CS:EIP: d8 4f 14 d9 5d f8 d9 ee d8 5d f8 d9 05 b8 36 a4 Stack dump: 0123aea4 00c6f660 00000300 00000400 00000000 00000000 0123bed0 0068efa0 60a42459 0068ef90 00c6f660 60b3096f 00c6f660 0068ef90 80000000 00000000 Talkback doesn't come up for anything, or I'd have sent talkback data. Sorry. If anyone has specific questions or if I can do anything to help give greater detail about this, please feel free to emails me anytime. Reproducible: Always Steps to Reproduce: 1. Put on crash helmet. 2. Start browser. 3. Crash. Actual Results: Boom. It crashes and I get a headache. Expected Results: Started. Again, Windows GPF data: MOZILLA caused an invalid page fault in module GKGFXWIN.DLL at 0177:60a41b34. Registers: EAX=00000000 CS=0177 EIP=60a41b34 EFLGS=00010246 EBX=0071fd40 SS=017f ESP=0068ef54 EBP=0068ef70 ECX=0068ef5c DS=017f ESI=00c6f660 FS=4de7 EDX=8000a728 ES=017f EDI=00000000 GS=0000 Bytes at CS:EIP: d8 4f 14 d9 5d f8 d9 ee d8 5d f8 d9 05 b8 36 a4 Stack dump: 0123aea4 00c6f660 00000300 00000400 00000000 00000000 0123bed0 0068efa0 60a42459 0068ef90 00c6f660 60b3096f 00c6f660 0068ef90 80000000 00000000
Comment 1•24 years ago
|
||
this worksforme with the 4/19 build under windows98 and NT.
Reporter | ||
Comment 2•24 years ago
|
||
I'm going to try tonight's nightly and see what happens.
Reporter | ||
Comment 3•24 years ago
|
||
Here's the crash data for tonight's (April 21, 2000) nightly: MOZILLA caused an invalid page fault in module GKGFXWIN.DLL at 0177:60af215c. Registers: EAX=00000000 CS=0177 EIP=60af215c EFLGS=00010246 EBX=00745e28 SS=017f ESP=0068f00c EBP=0068f028 ECX=0068f014 DS=017f ESI=0107aad0 FS=37ef EDX=00000001 ES=017f EDI=00000000 GS=0000 Bytes at CS:EIP: d8 4f 14 d9 5d f8 d9 ee d8 5d f8 d9 05 c8 46 af Stack dump: 01141b54 0107aad0 00000300 00000400 00000000 00000000 0111bc70 0068f058 60af2a81 0068f044 80000000 60be0b1a 0107aad0 0068f044 00000000 00000000 Any ideas on how I might be able to narrow this down? I just now, seconds ago, fired up M14 with crypto, and it worked perfectly. Started up quite fast actually. Maybe 10 seconds or so... Oh, and I'm running an Athlon 700, 128MB, Voodoo3 PCI card.
Comment 6•24 years ago
|
||
starsaa@usa.net and jblaine@shore.net I added you to the cc: list on this bug. can any of you get a nightly talkback build and try this again. if you are using a talkback-enabled mozilla build (such as): ftp://ftp.mozilla.org/pub/mozilla/nightly/latest/mozilla-win32-talkback.zip and submit the talkback report with your email address entered into it, then we can dig up the stack trace from there. Update the bug report if you submit a report from one of these crashes. Thanks.
Severity: blocker → critical
Comment 7•24 years ago
|
||
Downloaded 4/20 nightly talkback build. Same problem. Crash happens before talkback stuff is "enabled" or whatever. I assume talkback stuff is supposed to catch crashes and allow me to send a dump to you folks. It never gets that far.
Comment 8•24 years ago
|
||
have any of you tried running mozila with command line options like -console, -ProfileManager, -installer, -mail, etc. to see if it produces different results. Have you tried teh Installer.exed builds, the zipped builds... Is it creating the profile directory or anything besides the mozregistry.dat?
Reporter | ||
Comment 9•24 years ago
|
||
Me too. I always DL the talkback versions, and I can't get TB to generate a report. I'll grab the most recent nightly and try again. Also, I'll try to get as much debug data as possible, but so far this is it... Apparently Mozilla doesn't load far enough to start whatever TB needs to start. Asadotzler: Now, nothing besides mozregistry.dat gets built no matter what switch I use. I tried those too. Sory for not mentioning it. Alos, I've been using the ZIPped builds. Tonight I'll try an Installer build, since that's something I have yet to do. :)
Comment 10•24 years ago
|
||
1. mozilla.exe -ProfileManager = Same page fault. 2. mozilla.exe -console = Same page fault with the following shown in the DOS window: stdout directed to dynamic console stderr directed to dynamic console Profile Manager : Profile Wizard and Manager activites : Begin Profile Manager : Command Line Options : Begin Profile Manager : Command Line Options : End has multiple monitor apis is 1 ****** number of sceens 2 WEBSHELL+ = 1 gShellCounter: 0 WEBSHELL+ = 2 gShellCounter: 1 calling loadpage... startPage:: newProfile1_1 gShellCounter: 2 got a request has multiple monitor apis is 1 *** found screen 11a2 has multiple monitor apis is 1 3. mozilla.exe -installer = Same page fault. 4. mozilla.exe -mail = Same page fault. As mentioned already at the beginning of this bug report, the Installer version does the same thing (page fault). The only remnant I see from running it is that mozregistry.dat is created. No profile (Users50) area is set up ever.
Reporter | ||
Comment 11•24 years ago
|
||
Same errors here, except for -Console switch. Here's the output from that: stdout directed to dynamic console stderr directed to dynamic console Profile Manager : Profile Wizard and Manager activites : Begin Profile Manager : Command Line Options : Begin Profile Manager : Command Line Options : End ProfileManager : GetProfileDir ProfileManager : GetProfileDir Profile Manager : Profile Wizard and Manager activites : End
Reporter | ||
Comment 12•24 years ago
|
||
Still can't get Talkback to catch the crash. Is there anyway to either manually initialize Talkback or gather the information manually with another program? Mozilla.exe -console stdout directed to dynamic console stderr directed to dynamic console Profile Manager : Profile Wizard and Manager activites : Begin Profile Manager : Command Line Options : Begin Profile Manager : Command Line Options : End has multiple monitor apis is 1 ****** number of sceens 2 WEBSHELL+ = 1 WEBSHELL+ = 2 calling loadpage... startPage:: newProfile1_1 got a request has multiple monitor apis is 1 *** found screen 11a2 has multiple monitor apis is 1 GPF Data: MOZILLA caused an invalid page fault in module GKGFXWIN.DLL at 0177:60ae2139. Registers: EAX=00000000 CS=0177 EIP=60ae2139 EFLGS=00010246 EBX=0070f468 SS=017f ESP=0068efec EBP=0068f008 ECX=0068eff4 DS=017f ESI=0110a9f0 FS=2adf EDX=00000400 ES=017f EDI=00000000 GS=0000 Bytes at CS:EIP: d8 4f 14 d9 5d fc d9 ee d8 5d fc d9 05 c8 46 ae Stack dump: 018b4544 0110a9f0 00000300 00000400 00000000 018b5040 00000000 0068f038 60ae2b0b 0068f024 80000000 60bd0bd1 0110a9f0 0068f024 00000000 00000000 Also tried the installer EXE for M15 and the same thing happens. Crash on startup. Here's the GPF data from that: MOZILLA caused an invalid page fault in module GKGFXWIN.DLL at 0177:60a41b34. Registers: EAX=00000000 CS=0177 EIP=60a41b34 EFLGS=00010246 EBX=0070c718 SS=017f ESP=0068ef74 EBP=0068ef90 ECX=0068ef7c DS=017f ESI=00c7f750 FS=4957 EDX=8000ae60 ES=017f EDI=00000000 GS=0000 Bytes at CS:EIP: d8 4f 14 d9 5d f8 d9 ee d8 5d f8 d9 05 b8 36 a4 Stack dump: 01204324 00c7f750 00000300 00000400 00000000 00000000 01205c20 0068efc0 60a42459 0068efb0 00c7f750 60b3096f 00c7f750 0068efb0 80000000 00000000
Reporter | ||
Comment 13•24 years ago
|
||
Ok, from what I'm seeing, both jblaine@shore.net and I both have dual monitors. Could this be the problem? Have any of you guys tried testing this on a multi monitor Win98 system? I'm running Win98 4.10.1998, not Win98 SE, with a Voodoo3 PCI and a Mystique 220 for a secondary display. Just trying to narrow the possibilities... Still can't get Talkback to catch anything...
Comment 14•24 years ago
|
||
Indeed, I have a dual monitor setup at work (running 98 SE). My cards are both TNT2s, the primary one is a TNT2 Ultra. Note that I am home for the weekend and won't be able to help on this until Monday when I am back at work.
Comment 15•24 years ago
|
||
I am also using a dual monitor setup. M15 gave me the following error MOZILLA caused an invalid page fault in module GKGFXWIN.DLL at 0167:60a41b34. Registers: EAX=00000000 CS=0167 EIP=60a41b34 EFLGS=00010246 EBX=0071e028 SS=016f ESP=0068ef74 EBP=0068ef90 ECX=0068ef7c DS=016f ESI=0148f750 FS=5de7 EDX=80006248 ES=016f EDI=00000000 GS=0000 Bytes at CS:EIP: d8 4f 14 d9 5d f8 d9 ee d8 5d f8 d9 05 b8 36 a4 Stack dump: 021635d4 0148f750 0000023c 00000320 00000000 00000000 02164ec0 0068efc0 60a42459 0068efb0 0148f750 60b3096f 0148f750 0068efb0 80000000 00000000
Reporter | ||
Comment 16•24 years ago
|
||
Ok guys. I disabled my second monitor, and WITHOUT RESTARTING THE COMPUTER, I ran Mozilla, Nightly from 4-20. It worked perfectly. I re-enabled the monitor, and CRASH. I disabled it again, it worked, re-enabled screen 2, CRASH. I'm not a developer, but a long time tech, and to me, I'd say this is a DEFINITE dual monitor bug for 98, all versions. I don't know if it happens in NT or W2k. This bug is real, and now reproduceable. Dual displays on Win98 cause M15 and later to crash. I'm not a Moz developer, so I'm not going to do it, but _I_ feel this should be remarked as confirmed. I've run M15 and every nightly since M15 was released (I never DLed a nightly before the 19)th, and now they all exhibit the same behavior. Crash on Dual monitor (verifyed through three different users) and FOR ME, works when the second screen is disabled, no reboot needed.
Comment 17•24 years ago
|
||
copying pinkerton on this one if it is a dual monitor problem then maybe he knows it.
Assignee | ||
Comment 18•24 years ago
|
||
i've heard of problems, but no one in the building has been able to help me find them. our only dual-monitor PC is hyatt's win2k box and his machine exhibits none of the problems. i need help here. my pc is NT4.
Reporter | ||
Comment 19•24 years ago
|
||
Hi again. Ok, i've now managed to reproduce it on three different w98 dual display systems. My Athlon 700 with Win98, a Pentium MMX with W98SE, and a K6-500 with W98SE. I'll see what I can do about trying it under NT. Pinkerton: If you have any Win98 machines that you can play with for half an hour, and a second PCI vid card and monitor you can 'borrow' for a bit, it's very easy to setup 98 with dual displays. Just plug the vid card into a PCI clot with a higher address line than your primary (it's trial and error, if the new card winds up booting to the BIOS screen, shut down and swap them). 98 will see it and ask you for drivers. Then, go to Display Properties and click the new monitor. 98 will walk you thourgh the rest. Is there anything else I can do to help with this? Anyway to get talkback info manually? I'm a fairly advanced tech head, but don't do a lot of development. But I'm not gunshy about picking bits out of the sky, so any suggestions you have so I can get you guys more data is OK by me. I'm just glad I'm finally able to contribute SOMETHING to the project... :)
Comment 20•24 years ago
|
||
OK, Confirming this bug and updating summary to reflect real issue.
Status: UNCONFIRMED → NEW
Ever confirmed: true
Summary: Splash screen, then crash in GKGFXWIN.DLL → dual monitors under win98 crashes in GKGFXWIN.DLL
Comment 21•24 years ago
|
||
Probably a dup of bug #36239 both cause a crash in module gkgfxwin.dll and occured when using Windows 98, this appears to be a win98 only problem as it works on hyatt's win2k box.
Comment 22•24 years ago
|
||
jrgm, pink, does this look like bug 36239 to you?
Reporter | ||
Comment 23•24 years ago
|
||
after reading these comments, and bug 26239, I tried swapping the position of my monitors. I normally run like this: Primary | Secondary but in bug 36239 the guy had them swapped, with primary on the right. So I switched positions (logically) and then ran Mozilla. Worked until I went to access a menu, then crash. Exactly as described in bug 36239. Again, I switched back to my normal config with primary of left, and re-ran Mozilla. Crash on splash, as everything here says. Looks like a dupe. Too bad talkback still wouldn't catch the crash in the other config though...
Reporter | ||
Comment 24•24 years ago
|
||
Ack! Ignore that first bug number reference, typo!
Reporter | ||
Comment 25•24 years ago
|
||
Summary and Update: All errors under Win98 1. Mozilla crashed during splash screen with Primary display on left. Mozilla starts on primary, but crashes in GKGFXWIN.DLL. 2. With Primary display on right, top, or bottom, Mozilla starts fine. Sidebar is collapseable with no error. All buttons to the left of the Bookmarks button (Home, Mozilla.org, Tinderbox, etc.) work fine. Bookmark button, and all Menu bar functions crash mozilla in GKGFXWIN.DLL. 3. Disabling secondary display before running Mozilla results in no crashes at all. All buttons and menus work fine. 4. Disabling secondary monitor, then running Mozilla, and then RE-ENABLING the secondary display, WHILE MOZILLA IS STILL RUNNING, results in NO CRASHES. No matter the orientation, Mozilla will run fine if the secondary display is enabled after Mozilla is up, and Moz is on Primary. 5. Take number 4, and place secondary display on right or bottom. Mozilla runs, buttons work, no crash. BUT menus appear on primary display. On the right side of Primary if Secondary is on right, or on the bottom if secondary is on bottom. (Netscape 4.7x does something similar, menus open to the left side when primary is on right and window is on secondary, but never crashes) If secondary is on top or left, then the window becomes unresponsive until you move it back to the primary. 6. I have me secondary display on the right, with a small offset in levels. My secondary is a few pixels lower than my primary. This is because I can 'throw' the mouse to the top left, and click to close a program when it's maximized. The mouse get's stuck in the corner, just like on a single display system. When I arrange both displays to be level, then Mozilla starts fine, but crashes when you access a menu. It only crashes on startup when the displays are not lined up perfectly. My observations lead me to believe this: GKGFXWIN.DLL is having problems interpreting non-standard window coordinates begin passed to it by Windows (I'd say the GDI, right?). It can't handle extended X or Y coords when starting (multi monitor setups) and thus crashes on menu access. With offset diaplays, Moz crashes on start up. It can handle extended positive X,Y coords fine when it's running, but always draw menus on the primary display's coords. Offset secondary coords crash GKGFXWIN.DLL. It doesn't recognize negative coords at all. Even though it's located in a -X,-Y based window. When the Moz window is half on a neg coord screen and half on a positive coord screen, the menus on the positive side work, the neg side ones don't. Thus, Negative coordinates must be viewed at completely invalid. I haven't done any heavy duty programming like this in a while, but I'm going to browse through the source to see if I can make any other recommendations. Any idea who would be in charge of the code that works with these types of interactions?
Reporter | ||
Comment 26•24 years ago
|
||
Finally got it to generate some TB data. Incident number TB9339762Y It's commented as RE: bug 36239, typo meant to say this one, but that one leads here. My bad...
Reporter | ||
Comment 27•24 years ago
|
||
Ok. More digging. I think I've found where the problems lie, source wise. More so, Bug 21942, which is marked as Verified and Fixed, as of M15, is in reality NOT fixed, but alive and well, and I'd say a direct relation to this bug. Probably this bug's older brother. Can the module owner get to gether with me, whoever he/she/it/they/them are? I'd like to swap ideas...
Assignee | ||
Comment 28•24 years ago
|
||
well, that would be me.
Comment 30•24 years ago
|
||
pinkerton@netscape.com - would you like to take this bug out of Browser-General, or assign it to yourself? :-) Gerv
Assignee | ||
Comment 32•24 years ago
|
||
*** Bug 36239 has been marked as a duplicate of this bug. ***
Assignee | ||
Updated•24 years ago
|
Status: NEW → ASSIGNED
Target Milestone: --- → M18
Assignee | ||
Comment 33•24 years ago
|
||
*** Bug 37547 has been marked as a duplicate of this bug. ***
Comment 34•24 years ago
|
||
*** Bug 36462 has been marked as a duplicate of this bug. ***
Comment 35•24 years ago
|
||
*** Bug 38356 has been marked as a duplicate of this bug. ***
Assignee | ||
Comment 36•24 years ago
|
||
moving back to my beta2 radar.
Component: Browser-General → XP Toolkit/Widgets
Target Milestone: M18 → M17
Assignee | ||
Comment 38•24 years ago
|
||
I feel just sick about this, but i made a 98 development machine and i cannot duplicate the problem. at all. it works fine with two monitors. marking WORKSFORME as it works on win98 and win2k.
Status: ASSIGNED → RESOLVED
Closed: 24 years ago
Resolution: --- → WORKSFORME
Reporter | ||
Comment 39•24 years ago
|
||
First, It's real. The problem is when Mozilla is accessing the coords to draw the menu, it crashes. That occurs only after Moz has been successfully run on the system, so it has made a mozregistry.dat, etc. Second, tell me your system specs. Since I'm getting almost nowhere fiddling with the code, I may as well use my mountain of equipment for testing purooses...
Comment 40•24 years ago
|
||
Occurs on my system as well. specs: Windows 98 (first edition) Primary monitor: ATI All-In-Wonder 128 at 1280x1024, 32bpp (19") Secondary monitor: ATI Rage 128 at 1024x768, 16bpp (15") Experienced everyone else's problems, including the "has multiple monitor apis is 1" end console line. I do get a different dump, however: "MOZILLA caused an invalid page fault in module GKGFXWIN.DLL at 018f:60a84296. Registers: EAX=00000000 CS=018f EIP=60a84296 EFLGS=00010246 EBX=00ff64f0 SS=0197 ESP=0068f70c EBP=0068f728 ECX=0068f718 DS=0197 ESI=00000000 FS=655f EDX=00000500 ES=0197 EDI=00000028 GS=0000 Bytes at CS:EIP: d8 4e 14 d9 5d fc d9 ee d8 5d fc d9 05 d8 66 a8 Stack dump: 00ff64f0 00fe5f60 00000000 00000400 00000500 01aea5c0 00000000 0068f748 60a84a24 0068f738 0068f7c8 00000000 00000000 00000000 00000000 0068f788" Occured with every nightly of the past week (when I first enabled dual monitors). Request reopen status from someone..
Assignee | ||
Comment 41•24 years ago
|
||
ok, well, i'll grant that this crashes. i don't doubt that. i cannot duplicate the problem. No one with a development environment that can help debug has stepped forward. I'm at a loss, and this won't get fixed unless someone can help me.
Status: RESOLVED → REOPENED
Resolution: WORKSFORME → ---
Target Milestone: M17 → M19
Assignee | ||
Updated•24 years ago
|
Status: REOPENED → ASSIGNED
Reporter | ||
Comment 42•24 years ago
|
||
I have Visual Studio 6 Enterprise edition (damn thing cost enough). What can I do? I'm rusty in C, but not green. I narrowed down the problems to the fixes for bugs 21942 and 32611, version 1.5 of nsScreenManagerWin.cpp, and the couple files that go with it, but I don't see that as being as helpful as I'd like to be. Let me know what you'd liek to see. I can let VC++ grab the debug data, save the workspace and send it to you if you'd like...
Comment 43•24 years ago
|
||
*** Bug 39620 has been marked as a duplicate of this bug. ***
Comment 44•24 years ago
|
||
I'd be glad to help kill this bug. I get talkback data when I bomb on my win2k system, I'll send it in on monday. I'll also do a build and see if I see anything obvious in the debugger when it crashes.
Reporter | ||
Comment 45•24 years ago
|
||
Alos, the problem is in the "fixes" for the bugs where the menus would appear in the primary monitor even when the Moz window was on a secondary screen. I think the numbers referenced in the source were bug 21942 and bug 32611. IIRC bug 32612 is the GTK version of this whole multi screen deal...
Comment 46•24 years ago
|
||
I have a Dell Inspiron 7000. Windows 98 second edition and get the same problem. I don't know if it will help but I made a drwatson log. I put it at http://users.solve.net/~chadp1/WATSON00.WLG
Comment 47•24 years ago
|
||
Just built mozilla and tested; unfortunately this crash only occurs in optimized builds. I was able to browse just fine with the debug build. Will try building optimized with symbols next...
Assignee | ||
Comment 48•24 years ago
|
||
ok, well i fixed two things w/in that routine, but i'm still guessing about what the real problem is. marking fixed, would like it if someone could verify by repulling nsScreenManagerWin.cpp who has the crash.
Status: ASSIGNED → RESOLVED
Closed: 24 years ago → 24 years ago
Resolution: --- → FIXED
Comment 49•24 years ago
|
||
haven't pulled the mentioned changes yet. some debugging reveals that the call to screen->GetRect from nsDeviceContextWin :: ComputeFullAreaUsingScreen is trashing the stack; the crash then occurs at outRect->y = NSToIntRound(y * mDevUnitsToAppUnits); couldn't figure out why the stack is getting trashed though. will try the latest as soon as cvs lets me have it.
Comment 50•24 years ago
|
||
Still crashes in the same spot with the new changes. Removing the call to GetRect and filling in outRect with the single-monitor values prevents the crash; this also results in the menus being drawn on the wrong screen.
Assignee | ||
Comment 51•24 years ago
|
||
fuck. ok, well, there's no way i can fix this myself since i don't have a single machine anywhere that replicates it. obviously something is wrong deep within nsDeviceContextWin::FindScreen() but i have no idea what since i can't step through it.
Comment 52•24 years ago
|
||
Found and fixed the bug. All the multi-monitor APIs that you GetProcAddress for need to be declared as WINAPI. Otherwise they will smash the stack on return due to the calling convention mismatch. I changed the 3 typedefs in nsScreenWin and nsScreenManagerWin to include WINAPI and it worked on my box. Somebody just needs to check these changes in for me, I'm still waiting on my cvs write access.
Assignee | ||
Comment 53•24 years ago
|
||
you rock! i'll do that today! see, this is what you get when you leave win32 programming to a mac weenie ;)
Assignee | ||
Comment 54•24 years ago
|
||
can you give me a patch? i can't seem to get that to compile. thanks.
Status: REOPENED → ASSIGNED
Comment 55•24 years ago
|
||
nsScreenWin.cpp typedef BOOL (WINAPI *GetMonitorInfoProc)(HMONITOR inMon, LPMONITORINFO ioInfo); nsScreenManagerWin.cpp typedef HMONITOR (WINAPI *MonitorFromRectProc)(LPCRECT inRect, DWORD inFlag); typedef BOOL (WINAPI *EnumDisplayMonitorsProc)(HDC, LPCRECT, MONITORENUMPROC, LPARAM);
Reporter | ||
Comment 56•24 years ago
|
||
First, THANK YOU! Second, YOU SUCK! I went to cash my paycheck, and stopped at a Borders to browse the Computer Programming section with a printout of the source to those files. And I found the exact thing you did! =-] Congrats, and thank you. And Pink? Why is it you can call yourself a Mac Weenie, but when we PeeCee nuts call you that, you get upset? =-] Also, one note I also found that may be of service in the future: For Multimon systems with displays that are offset (like mine were) rather than forming one continuous screen, while Windows prevents the mouse from going off the display areas, there are addressable areas of the virtual display that are accessable to programs (it's a feature, not a bug, really). Should folks who wish to use the Gecko core in other projects should make note to specifically address the displays (like this is doing) rather than the entire virtual screen.
Comment 57•24 years ago
|
||
Left as an exercise to the reader: why did this only show up in optimized builds? The stack would still have been off by 8 bytes following the return from GetMonitorInfo, I guess maybe there is enough padding in debug builds to allow that to go unpunished.
Comment 58•24 years ago
|
||
*** Bug 40423 has been marked as a duplicate of this bug. ***
Assignee | ||
Comment 59•24 years ago
|
||
fixed, oh finally fixed.
Status: ASSIGNED → RESOLVED
Closed: 24 years ago → 24 years ago
Resolution: --- → FIXED
Comment 60•24 years ago
|
||
Can the folks who experienced this crash let us know if you're running OK with tomorrow's win32 build. Thanks, and a big thanks to juberti@aol.com and jesusx@who.net (jesusx) for tracking this down!
Whiteboard: [nsbeta2+] 6/1 → [nsbeta2+] 6/1 (needs external verification)
Reporter | ||
Comment 61•24 years ago
|
||
I've been pulling down builds all day, but they're not working yet. As soon as I get a functioning build, you'll know. I'm glad to have helped (and look forward to trying to squash more bugs) and absolutely love juberti@aol.com for finding the fix. =-]
Comment 62•24 years ago
|
||
The fix won't be in the nightlies until the next set that comes out (tomorrow a.m. PST) -- mike just checked the code in about a half-hour ago. You can of course roll your own though, and you'll get the fix now.
Reporter | ||
Comment 63•24 years ago
|
||
IT WORKS! IT WORKS! IT REALLY REALLY WORKS!!! It's fast, friendly, and now the official browser of this egomaniac! Woohoo! Oh, and the fix seems to have solved our problem. Good work fellows. Cherrio... =-]
Reporter | ||
Comment 64•24 years ago
|
||
Last Comment (posted with Build 2000052420!) First, the Name's Grey Hodge. Yessiree. And tomorrow, I'll be filing a flurry of bugs, after I DL another build to bang around and verify first. Second, this fix, as simple as it was, sows the great amount of time and effort in Mozilla because Mozilla, even with one window spread across two displays, performs MUCH faster than even any Explorer window (NOT IE, any ol' explorer window), even when resizing! Blistering speed. In short, and closing, Moz seems to now work _flawlessly_ with multiple displays (at least on the two I have). <knock wood>
Comment 66•24 years ago
|
||
Works great now, thanks for the good work!
You need to log in
before you can comment on or make changes to this bug.
Description
•