January 2003 - Wine-Devel - list.winehq.org

Re: Patch for Wine-20030115 heap allocation on Solaris 8 x86
by John Wehle Jan. 18, 2003

Jan. 18, 2003

>On Fri, Jan 17, 2003 at 01:52:08PM -0500, John Wehle wrote: >> Currently wine exits after complaining about an invalid heap during >> startup on Solaris 8 x86. What happens is mmap is returning an address >> above 0xc0000000 which HEAP_InitSubHeap passes to NtAllocateVirtualMemory >> who rejects it. Perhaps someone familiar with the code can suggest >> a better solution or explain why the ADDRESS_SPACE_LIMIT check is there. > Correct me if I'm wrong, but AFAICS this is even the third time that you > submit this. Yep. > Don't you think that it would have been time to ask Alexandre > what the hell he's been thinking about constantly rejecting it ? ;-) Didn't know it had been rejected. I'm really not up on how wine development is done. In general I'm suppose to email julliard(a)winehq.com if I don't receive an ACK / NACK regarding a patch? -- John ------------------------------------------------------------------------- | Feith Systems | Voice: 1-215-646-8000 | Email: john(a)feith.com | | John Wehle | Fax: 1-215-540-5495 | | -------------------------------------------------------------------------

2 1

Check for block device in DRIVE_ReadSuperblock
by Michal Janusz Miroslaw Jan. 18, 2003

Jan. 18, 2003

I thought, that this may be useful to somebody ... If this patch were here before it would save me a few hours searching why every program hanged on commdlg open/save file dialogs ... I'm sending it here because I'm not sure whether I should be more/less paranoid and stat() the file before open() or not, or even if this check really should go into this function. Decide. ;) Changelog: [MQ] Add check for block device in DRIVE_ReadSuperblock Index: files/drive.c =================================================================== RCS file: /home/wine/wine/files/drive.c,v retrieving revision 1.83 diff -u -r1.83 drive.c --- files/drive.c 13 Jan 2003 20:44:13 -0000 1.83 +++ files/drive.c 18 Jan 2003 02:05:38 -0000 @@ -623,20 +623,36 @@ int fd; off_t offs; int ret = 0; + struct stat stat_buf; if (memset(buff,0,DRIVE_SUPER)!=buff) return -1; if ((fd=open(DOSDrives[drive].device,O_RDONLY)) == -1) { - struct stat st; if (!DOSDrives[drive].device) ERR("No device configured for drive %c: !\n", 'A'+drive); else ERR("Couldn't open device '%s' for drive %c: ! (%s)\n", DOSDrives[drive].device, 'A'+drive, - (stat(DOSDrives[drive].device, &st)) ? + (stat(DOSDrives[drive].device, &stat_buf)) ? "not available or symlink not valid ?" : "no permission"); ERR("Can't read drive volume info ! Either pre-set it or make sure the device to read it from is accessible !\n"); PROFILE_UsageWineIni(); return -1; + } + + /* MQ: I guess, we don't have to merge this call with the one above - some weird user + may try to swap the file before stat() and open() while we're not watching ;) + OTOH if we open FIFO we're stuck anyway, so FIXME */ + if (fstat(fd, &stat_buf) < 0) { /* shouldn't happen since we just opened that file */ + ERR("fstat() failed for opened device '%s' (drive %c:) ! IT SHOULDN'T HAPPEN !!!\n", + DOSDrives[drive].device, 'A'+drive); + ret = -1; + goto the_end; + } + if (!S_ISBLK(stat_buf.st_mode)) { + ERR("Device '%s' (drive %c:) is not a block device - check your config\n", + DOSDrives[drive].device, 'A'+drive); + ret = -1; + goto the_end; } switch(DOSDrives[drive].type)

1 0

Heap* in memory/heap.c
by Michal Janusz Miroslaw Jan. 18, 2003

Jan. 18, 2003

While checking why my new shiny man pages for HeapAlloc and friends didn't get generated, I found something weird. The functions HeapAlloc, HeapFree, (...) are redirected to ntdll equivalents RtlAllocateHeap, RtlFreeHeap, (...). There are more functions that are of return RtlXXX(args); type but are not forwarded in kernel32.spec. My questions: * Should the forwardings be added for the rest of functions ? * What about documenting them (and changing c2man.pl to support it) ? -- Michal Miroslaw

1 0

Getting a program to crash
by Duane Clark Jan. 17, 2003

Jan. 17, 2003

Howdy, I am trying to make a program crash ;) Instead, the program apparently captures exceptions, pops up a dialog, and continues on. I am hoping for a way to prevent that from happening, and instead falling into the debugger. It does call EXC_RtlRaiseException when the exception occurs, but it is not clear to me how that function works. Hopefully I can comment out something there to get it to crash properly?

2 1

wine-releases has gone?
by Mike Hearn Jan. 17, 2003

Jan. 17, 2003

Hi, On the Wine release notice it talks about a mailing list for large diffs between the releases, which would be useful for me (no cvs access at the moment at work), but the list appears to be dead, with no archives available. Is it still in use? -- Mike Hearn <m.hearn(a)signal.qinetiq.com> QinetiQ - Malvern Technology Center

3 2

Re: Patch for Wine-20030115 heap allocation on Solaris 8 x86
by Andreas Mohr Jan. 17, 2003

Jan. 17, 2003

On Fri, Jan 17, 2003 at 01:52:08PM -0500, John Wehle wrote: > Currently wine exits after complaining about an invalid heap during > startup on Solaris 8 x86. What happens is mmap is returning an address > above 0xc0000000 which HEAP_InitSubHeap passes to NtAllocateVirtualMemory > who rejects it. Perhaps someone familar with the code can suggest > a better solution or explain why the ADDRESS_SPACE_LIMIT check is there. Correct me if I'm wrong, but AFAICS this is even the third time that you submit this. Don't you think that it would have been time to ask Alexandre what the hell he's been thinking about constantly rejecting it ? ;-) Have a nice day ! -- Andreas Mohr Stauferstr. 6, D-71272 Renningen, Germany Tel. +49 7159 800604 http://mohr.de.tt

2 1

wine robustness?
by Dan Kegel Jan. 17, 2003

Jan. 17, 2003

Say, how much address space protection is there in Wine? If a Wine process follows a wild pointer or passes garbage arguments to a Windows function call, are Wineserver and the other wine processes protected? Imagine a user running Word, and also some less reliable windows program. When that other program crashes, it shouldn't take Word with it. One way to check that kind of protection is with programs like 'crashme': http://people.delphiforums.com/gjc/crashme.html Has anyone tried running crashme on wine yet? - Dan -- Dan Kegel http://www.kegel.com http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=78045

2 1

Away for 9 days
by Dimitrie O. Paun Jan. 17, 2003

Jan. 17, 2003

Hi folks, I'll be away for 9 days starting tomorrow, skiing. Just in case you wonder whether I've fallen off the net. :) -- Dimi.

2 1

Re: Patch for Wine-20030115 heap allocation on Solaris 8 x86
by Dan Kegel Jan. 17, 2003

Jan. 17, 2003

John Wehle wrote: > Currently wine exits after complaining about an invalid heap during > startup on Solaris 8 x86. What happens is mmap is returning an address > above 0xc0000000 which HEAP_InitSubHeap passes to NtAllocateVirtualMemory > who rejects it. Perhaps someone familar with the code can suggest > a better solution or explain why the ADDRESS_SPACE_LIMIT check is there. See http://groups.google.com/groups?selm=%2Bjq*-eaVn%40news.virata.com Not that I know, but I suspect Wine is trying to avoid handing out pointers that Windows can't possibly hand out. I have a strong feeling of deja vu on this question... has it never been answered before? - Dan -- Dan Kegel http://www.kegel.com http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=78045

1 0

system call tracing
by Dan Kegel Jan. 17, 2003

Jan. 17, 2003

This looks like an interesting tool: http://razor.bindview.com/tools/desc/strace_readme.html It only traces NT native calls, but what the hey, it comes with source. -- Dan Kegel http://www.kegel.com http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=78045

1 0