June 2006 - Wine-Devel - list.winehq.org

Re: Wine, Darwine, CodeWeavers-- Intel MacOS X

by James Hawkins

On 6/26/06, William Knop <william.knop(a)gmail.com> wrote: > I think you missed my point. I know people are working on it. The > problem is "a few people, including Codeweavers, are working on it," > which is just about the sum-total of the information out there, is not > enough. There is very little information out there for potential > developers to know what's going on with wine and intel macos x. For > instance, a status page describing to-do items would be a start (I'm > going to add it to the wine wiki when I figure out what the status > is). It should be mentioned on winehq, so macos x developers know to > investigate further when they check wine out. Expecting people to dig > through the code or the mailing list is unreasonable. Is this really > only obvious to me? > You could just ask, "Hey, what's the status of Wine on Intel Max OS X?". No one has really stepped up to work on that aspect of Wine, outside of Codeweavers, so there's not a whole lot of incentive to put this information up. So if you're wanting to work on this, step up to the plate, and I'm sure we'll be more than willing to help. -- James Hawkins

June 26, 2006

Re: Wine, Darwine, CodeWeavers-- Intel MacOS X

by William Knop

I think you missed my point. I know people are working on it. The problem is "a few people, including Codeweavers, are working on it," which is just about the sum-total of the information out there, is not enough. There is very little information out there for potential developers to know what's going on with wine and intel macos x. For instance, a status page describing to-do items would be a start (I'm going to add it to the wine wiki when I figure out what the status is). It should be mentioned on winehq, so macos x developers know to investigate further when they check wine out. Expecting people to dig through the code or the mailing list is unreasonable. Is this really only obvious to me? I don't mean to be rude; I just see some easy ways to attract mac people to wine. Wine is likely to be an important addition to mac users' software base in the future. So many people want the "mac experience," but have to run windows software. Many more people, I'd say, than those who wanted the mac experience and the unix software suite, and look at how popular fink is. It's a disservice to both parties to not properly advertise. On 6/25/06, Dan Kegel <dank(a)kegel.com> wrote: > William Knop writes: > > What's the story with wine on intel os x? > > A few people, including Codeweavers, are working on it. > More would be welcome. There is no politics keeping > people away, as far as I know. If you run into any developer > who says that politics is keeping him or her from > submitting patches to wine-patches, please let me know. > - Dan > > >

June 26, 2006

Re: winspool.drv: Write-strings warnings fix

by Andrew Talbot

Mike McCormack wrote: > These strings should really be const. IMO, a better fix would be to add > a cast to cast them back to (char*) than to change their storage class. > The trouble is, if you leave them as they were, you get a write-strings warning, since you are pointing a writeable pointer at a read-only string). But if you cast, thus: CHAR *dtype = (LPSTR) "RAW"; ... then you swap the write-strings warning for a cast-qual one. In general, it's a solution that Alexandre has accepted. -- Andy.

June 26, 2006

Re: Very large files

by Damjan Jovanovic

--- John Willis <jswillis93105(a)verizon.net> wrote: > Howdy, > > I have a large VS 6.0 C++ program and I'm running it > under WINE under Redhat Linux. > Everything works fine, but it doesn't handle file > sizes greater than 4gb like it can running > directly under Windows XP. > I've tweaked everything that I can find that seems > to deal with file sizes. How do you access the file? With stdio streams (fopen(), fread(), ...), MSVCRT's open()/read(), or with KERNEL32's CreateFile(), ReadFile(), ... ? > Is there a problem with WINE itself handling very > large file sizes? Which version of the Linux kernel and which filesystem are you using? I'll run some tests at home and see whether it works for me. > Thanks in advance, > John> > __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

June 26, 2006

Re: [AppDb] [2/3] safe functions

by Jonathan Ernst

Le dimanche 25 juin 2006 à 20:00 -0400, Chris Morgan a écrit : > Hi Jonathan. > > You'll want to talk to EA about the filtering changes. The plan is to > filter using the same syntax and flags that the php filter extension > is going to use so we can easily switch over to this extension in the > future. I know we could use PEAR and we could also use a database abstraction layer, I just thought my solution was better because it has proven to work well on several projects I worked recently and is recommanded by the php manual (and it makes queries more readable than using other syntaxes). > > Also, I've submitted a patch for review to appdb(a)winehq.com and > wine-patches(a)winehq.com that removes all of our get_magic_quotes_gpc() > use and adds a check in include/incl.php that warns and prevents appdb > from running if magic quotes is enabled. So you shouldn't need to > have any get_magic_quotes_gpc() checks anymore. Isn't it better to support both configurations ? My solution works with or without magic quotes. > > I also noticed your quote_smart_sql() call. This call isn't used > anywhere, we shouldn't add calls to functions that aren't called. We It is used in 3/3. > also already have a function that will make sql calls safe called > query_paramters() in include/db.php. Also, do we want to strip tags > from sql? Won't that remove all tags from things like app/version > descriptions, comments and notes? No, there is a parameter in this function (quote_smart_sql). By default we don't remove html, but for some fields we might want to filter out html (comment titles, etc.) Thanks. Jonathan

June 26, 2006

Re: [AppDB] - protect against sql injection in select, update and delete statements

by Tony Lambregts

Chris Morgan wrote: >> Oh for crying out loud. Go ahead and automate if you know how to. If you or >> someone else can show me how I am willing to work on it too, I an not >> opposed to automating all of the testing if that is possible. >> >> In the mean time am against these large patches that are difficult to test >> because the are so large. >> >> > > The difficulty isn't that a particular change is large, although yes, if only > a few lines of code that were only called from a single location were changed > this would make it easy to test. The issue is that the appdb is so large and > complex that we aren't using time efficiently by testing manually. > I'm not arguing against automated testing, however not having automated testing is not an excuse for not testing. > Automated testing isn't all that difficult to implement. We can start out > with tests for classes, make sure we can create a new user, change the users > password and other info and delete the user. Test creating applications and > versions. We should even be able to fill in form data and simulate the user > entering data and clicking on the submit button. > This is a fine idea... > I already have many of the tests for the user class completed from this last > October. I would like to see this. > > Let me finish up closing these sql holes and I'll clean the tests > up and submit them as a basis for our automated testing. > I'll repeat I do not want patches going into the live system without being tested. Break them up so that they are easier to test. For the life of me I do not see how on earth you think is this an unreasonable request. I really do not enjoy arguing with you about this but I feel I am forced too because testing is the only way I can see to ensure that we don't keep busting the AppDB. I have said it before and I will say it again we have lost more data so far through bad patches then through security breaches. -- Tony Lambregts

June 26, 2006

Re: [AppDB] - protect against sql injection in select, update and delete statements

by Chris Morgan

> Oh for crying out loud. Go ahead and automate if you know how to. If you or > someone else can show me how I am willing to work on it too, I an not > opposed to automating all of the testing if that is possible. > > In the mean time am against these large patches that are difficult to test > because the are so large. > The difficulty isn't that a particular change is large, although yes, if only a few lines of code that were only called from a single location were changed this would make it easy to test. The issue is that the appdb is so large and complex that we aren't using time efficiently by testing manually. Automated testing isn't all that difficult to implement. We can start out with tests for classes, make sure we can create a new user, change the users password and other info and delete the user. Test creating applications and versions. We should even be able to fill in form data and simulate the user entering data and clicking on the submit button. I already have many of the tests for the user class completed from this last October. Let me finish up closing these sql holes and I'll clean the tests up and submit them as a basis for our automated testing. Chris

June 26, 2006

Re: [AppDB] - protect against sql injection in select, update and delete statements

by Tony Lambregts

Chris Morgan wrote: > On Sunday 25 June 2006 10:00 pm, Tony Lambregts wrote: >> Chris Morgan wrote: >>> Testing once beats testing over and over and over again IMO. >> Thats why you want to make the patches small. Small patches are easy to >> review, test and prove the correctness of. Big patches are difficult to >> test and one bug rejects the whole thing, so you have to start all over >> again. >> >>> I suspect that if your testing looks good and mine does as well then we >>> should be ready to go. >> I will NOT be able to test a big patch again for a while, nor am I really >> willing to. I took quite a while out of my weekend to do the testing >> because I don't want a bad patch going into production again. We have had >> enough bad feelings going on over this that it was easier for me to do the >> testing then to continue arguing with you. I really have a limit though and >> I ask you politely to meet me half way by breaking it up. >> >>> I'll fix up moving test results, that should be an easy one, and do >>> another once over of the patch before comitting it. >>> I'm pretty confident that it is all set, I've tested submitting notes, >>> comments, applications, distributions etc here and it looks good. >> I do not support that idea. If you break the patch up into smaller chunks >> then Those can be tested and hopfully put in. >> >> -- >> >> Tony Lambregts > > I'll break up the patch if you'll start submitting patches that implement > automated testing. > > Otherwise our testing is always going to have poor coverage and is going to be > a waste of time. It might be more difficult to implement automated testing > but it will improve the codebase by allowing us to test it quickly and > reproducably. > > Chris > > Oh for crying out loud. Go ahead and automate if you know how to. If you or someone else can show me how I am willing to work on it too, I an not opposed to automating all of the testing if that is possible. In the mean time am against these large patches that are difficult to test because the are so large. -- Tony Lambregts

June 26, 2006

Re: winspool.drv: Write-strings warnings fix

by Mike McCormack

Andrew Talbot wrote: > + static CHAR dtype[] = "RAW", > + processor[] = "WinPrint", These strings should really be const. IMO, a better fix would be to add a cast to cast them back to (char*) than to change their storage class. Mike

June 26, 2006

Re: [AppDB] - protect against sql injection in select, update and delete statements

by Chris Morgan

On Sunday 25 June 2006 10:00 pm, Tony Lambregts wrote: > Chris Morgan wrote: > > Testing once beats testing over and over and over again IMO. > > Thats why you want to make the patches small. Small patches are easy to > review, test and prove the correctness of. Big patches are difficult to > test and one bug rejects the whole thing, so you have to start all over > again. > > > I suspect that if your testing looks good and mine does as well then we > > should be ready to go. > > I will NOT be able to test a big patch again for a while, nor am I really > willing to. I took quite a while out of my weekend to do the testing > because I don't want a bad patch going into production again. We have had > enough bad feelings going on over this that it was easier for me to do the > testing then to continue arguing with you. I really have a limit though and > I ask you politely to meet me half way by breaking it up. > > > I'll fix up moving test results, that should be an easy one, and do > > another once over of the patch before comitting it. > > I'm pretty confident that it is all set, I've tested submitting notes, > > comments, applications, distributions etc here and it looks good. > > I do not support that idea. If you break the patch up into smaller chunks > then Those can be tested and hopfully put in. > > -- > > Tony Lambregts I'll break up the patch if you'll start submitting patches that implement automated testing. Otherwise our testing is always going to have poor coverage and is going to be a waste of time. It might be more difficult to implement automated testing but it will improve the codebase by allowing us to test it quickly and reproducably. Chris

June 26, 2006

Re: [AppDB] - protect against sql injection in select, update and delete statements

by Tony Lambregts

Chris Morgan wrote: > Testing once beats testing over and over and over again IMO. Thats why you want to make the patches small. Small patches are easy to review, test and prove the correctness of. Big patches are difficult to test and one bug rejects the whole thing, so you have to start all over again. > I suspect that if your testing looks good and mine does as well then we > should be ready to go. I will NOT be able to test a big patch again for a while, nor am I really willing to. I took quite a while out of my weekend to do the testing because I don't want a bad patch going into production again. We have had enough bad feelings going on over this that it was easier for me to do the testing then to continue arguing with you. I really have a limit though and I ask you politely to meet me half way by breaking it up. > > I'll fix up moving test results, that should be an easy one, and do > another once over of the patch before comitting it. > I'm pretty confident that it is all set, I've tested submitting notes, comments, > applications, distributions etc here and it looks good. > I do not support that idea. If you break the patch up into smaller chunks then Those can be tested and hopfully put in. -- Tony Lambregts

June 26, 2006

re: Wine, Darwine, CodeWeavers-- Intel MacOS X

by Dan Kegel

William Knopp writes: > What's the story with wine on intel os x? A few people, including Codeweavers, are working on it. More would be welcome. There is no politics keeping people away, as far as I know. If you run into any developer who says that politics is keeping him or her from submitting patches to wine-patches, please let me know. - Dan

June 26, 2006

Re: [AppDb] [3/3] Comments handling cleanup

by Chris Morgan

The makeSafe() changes for filtering data and the query_parameters() changes for sql injection parameters are related but independent changes. It seems like query_parameters() is a better fix than putting inline sprintf()s and quote_safe_sql() calls. query_parameters() encapsulates the calls to whatever 'escape' function that we choose. query_parameters() also uses the syntax that pear db uses for place holders, ?, ~ and &. Chris On 6/25/06, Jonathan Ernst <jonathan(a)ernstfamily.ch> wrote: > Hi, > > Here is a cleaned up version of the comments handling that better fixes > (imho) all problems related to sql and html injection (it requires the previous patches to be applied). > > It is based on php manual's best practices for avoiding injection. > > I'd be very glad if we'll use such mechanism for the rest of the > queries and get rid of compile_whatever, makeSafe & co. > > Changelog: > - avoid sql/html injection in the comments > > Files changed: > - addcomment.php > - deletecomment.php > - include/comment.php > > > > -----BEGIN PGP SIGNATURE----- > Version: GnuPG v1.4.2.2 (GNU/Linux) > > iD8DBQBEnt7XIW5mR/h6b38RAqtqAKCw7qX+8JTVDI0VvvHSfZTbeGYJOQCglZtg > gRVPintjJeZ1yhupSF5V+RE= > =6bPd > -----END PGP SIGNATURE----- > > > > > > >

June 26, 2006

Re: [AppDb] stuff thats broken

by Chris Morgan

Oh. I had no idea that you had re-enabled makeSafe(). You almost certainly do not want to do that, it was disabled because it was broken. makeSafe() likely will never use addSlashes() since that protection was intended to make variables safe for use with sql. Future implementations of makeSafe() will instead perform some magic like: $iNumeric = makeSafe($iNumeric, FILTER_NUMERIC); Chris On 6/25/06, Tony Lambregts <tony.lambregts(a)gmail.com> wrote: > Chris Morgan wrote: > >>> 5.) If you add a hyperlink in a note it is broken with makesafe() > >>> turned on. > >> I'm not sure this has anything to do with makeSafe(). Right now makeSafe() > >> is effectively disabled and I've carefully reviewed the changes to > >> addAppNode.php but I can't see how any of the changes broke notes. > >> > >> On my local machine after previewing a note I'm seeing any url I put in > >> prefixed by the current page url. On the appdb when previewing everything > >> appears to be working correctly. > >> > >> I think this may be a legit bug with the injection changes and I'll fix it > >> before committing. > >> > > > > Just tested again here with the injection changes and hypertext(tested with a > > url) in notes looks good to me. > > > > Chris > > > > > > > > let me clarify that this is with makeSafe() turned on. IE male safe defined as this: > > function makeSafe($var) > { > $var = trim(addslashes($var)); > return $var; > } > > This has nothing to do with magic_quotes. I have applied your > magic_quotes_warning.patch and do not get the warning message that I should if > get_magic_quotes_gpc() was on. I have this in my /etc/php.ini > > ; Magic quotes for incoming GET/POST/Cookie data. > magic_quotes_gpc = Off > > The reason I have makeSafe() "turned on" is that I assume that eventually we > will make it use addslashes() in it and I wanted to find any issues that doing > so would produce in the course of testing. > > What I have found is that input fields that use xinha have this issue when > makeSafe() uses addslashes(). > > Does that help... > > PS: I am still not finished testing the select/update/delete patch > > -- > > Tony lambregts > > > > >

June 26, 2006

Re: [AppDb] [2/3] safe functions

by Chris Morgan

Hi Jonathan. You'll want to talk to EA about the filtering changes. The plan is to filter using the same syntax and flags that the php filter extension is going to use so we can easily switch over to this extension in the future. Also, I've submitted a patch for review to appdb(a)winehq.com and wine-patches(a)winehq.com that removes all of our get_magic_quotes_gpc() use and adds a check in include/incl.php that warns and prevents appdb from running if magic quotes is enabled. So you shouldn't need to have any get_magic_quotes_gpc() checks anymore. I also noticed your quote_smart_sql() call. This call isn't used anywhere, we shouldn't add calls to functions that aren't called. We also already have a function that will make sql calls safe called query_paramters() in include/db.php. Also, do we want to strip tags from sql? Won't that remove all tags from things like app/version descriptions, comments and notes? Chris On 6/25/06, Jonathan Ernst <jonathan(a)ernstfamily.ch> wrote: > Hi, > > The code I made is not dependent on magic_quote value and shouldn't > allow anybody to mess with the values read or written from/to the > database. Also it doesn't require to make copies of every single string the user > passes to us. > > It is based on php manual's best practices for avoiding injection. > > I'd be very glad if we'll use such mechanism for the rest of the > queries and get rid of compile_whatever, makeSafe & co. > > Changelog: > - functions to avoid sql/html injection without having issues with addslashes/magic_quotes_gpc > > Files changed: > - include/util.php > > > -----BEGIN PGP SIGNATURE----- > Version: GnuPG v1.4.2.2 (GNU/Linux) > > iD8DBQBEnt7SIW5mR/h6b38RAii/AKCB26o0DAM978gQz8baVz6VuubW3ACgqBoJ > Or4cDg49YqCUD3dughsW1oY= > =keDV > -----END PGP SIGNATURE----- > > > > > > >

June 26, 2006

Wine, Darwine, CodeWeavers-- Intel MacOS X

by William Knop

Hello all, What's the story with wine on intel os x? It's barely mentioned in the wine wiki and on the darwine website (not at all on winehq). I'm under the impression darwine is concentrating on running wine on ppc/ darwin. I know codeweavers is working specifically on making wine on intel/darwin happen, however there is no status page anywhere. Just from keeping up with the mailing list, I've gleaned that Quartz and Core Audio are being worked on, in addition to a stack alignment patch. They seem to be pretty far along, but I don't really know. Are there any other intel/darwin areas that require attention or are being worked on? I'd like to add this info to the wine wiki. It seems odd that intel os x is sort of not officially recognized by wine, what with there being no mention of it on the download page, to do lists, or bugzilla... I know there're some politics involved, but it really should have more attention. Its existence is practically not even recognized by wine. At the least, that's turning away potential wine developers, and the thing is, the mac community could be all about wine with a little effort. Thanks, Will

June 25, 2006

Re: SecuRom investigation

by Martin Owen

Currently it seems to create drives based on directory locations, but there is no way of specifying what kind of drive it appears to be to windows. perhaps this would help games identify the cdrom drive and some games install of multiple discs. I can't help but feel wine has no support for cd drives and depends on the file system in linux to do the leg work. On 6/25/06, n0dalus <n0dalus+wine(a)gmail.com> wrote: > On 6/25/06, Martin Owen <doctormo(a)gmail.com> wrote: > > > > which if we could get our hands on the source code would at least give us > > more information about how this security works and may even allow us to work > > around it perminently with in wine. > > > > One easy way of improving compatibility with copy protections without > any source code is to make all drives visible to wine applications > appear to be SCSI. The main copy protections don't "work" on SCSI > drives so the games are allowed to run unchecked. I'm not sure how > hard this would be to setup in wine at the moment though. > > n0dalus. >

June 25, 2006

Opengl states in wined3d

by Stefan Dösinger

Hello, The way opengl states are managed in wined3d at the moment is a bit messy and inefficient. Basically it is a mixture of modifying the opengl settings when the application modifies the d3d settings and brute-force applying the rest in drawprim. The situation has got worse with the addition of the DirectDraw blitting code and it is time to clean that up and make it more efficient. Basically during primitive drawing the opengl setup must equal the one requested by the application. Some things can be translated 1:1, e.g. WINED3DRS_LIGHTINGENABLE to glEnable(GL_LIGHTING). Other things like fog or texture stage states are more complex. Some operations in wined3d, for example unlocking a render target or doing a 2D blt require special opengl settings. At the moment we have the following situation: * Some things are applied when the app requests a change * Some parts are applied by brute force in drawPrimitive * BltOverride and unlockRect read the old gl states, modify it the way they need it and restore the old states back. This has some problems: * Setting some things during each drawprim call wastes quite a few resources, due to the loops that and gl calls. * If an application performs only blts(e.g. 2D only app) or only locks the render target(playing a movie) storing and restoring the gl state is a waste of resources * Brute force applying opengl states is likely to re-set the old state unnecessarily. Opengl does not assure that redundant calls are cheap. * if an application sets and resets a state for some reason between 2 drawprim calls that involves 2 not really necessary opengl state switches * Anything else I didn't think of? I can think of a few ways to solve this: * Do not do any opengl changes in SetRenderState, and apply all states in drawprim. This way UnlockRect and Blt don't have to care for resetting the things they changed and redundant changes can be catched nicely * Apply everything when the app requests to do so, and take as much things out of drawprim. This keeps drawprim small and efficient. * Use a mixed style like it is done with transformed vs untransformed vertex drawing with last_was_rhw. That makes it easy to find out if reapplying anything is needed and frees other functions from resetting everything. To avoid re-setting an old setting again the current opengl state could be stored in the d3ddevice. I think this should be done in any case, and depending on the nature of some parameters the opengl state or the d3d state should be kept in there(or both). So we shouldn't use a d3d stateblock but instead our own structure where we can add the stuff we need. I also think we should try to get rid of as much things as possible in drawprim. I do not mean entirely removing it, but having a simple flag which tells if a bigger number of states needs attention, like last_was_rhw does. I think about adding a last_was_blit flag which is set in BltOverride and UnlockRect. I'm not sure about the texture stage states and sampler states. In d3d the settings seem to be per-stage, while in opengl I think they are per texture. If so we could store the last d3d settings that the texture was used with in the texture too see if we have to reapply them. A partially related sidenote about drawStridedSlow: It has a loop which iterates through the vertices to draw, and in this loop there are if statements checking which data the vertex contains. So if some data isn't there this is at worst a comparison + a jump. I had a look at the vertex data 3DMark2000 uses and removed checks and handling for the things it didn't use for testing. 3DMark2000 doesn't need drawStridedSlow a lot, but yet the score increased from 4879 to 5013 3DMarks. As a comparison forcing drawStridedFast gets about 5500 3dmarks. The fps in the low detail helicopter test increased from 95.6 to 99.2(105 with drawStridedFast). Cheers, Stefan

June 25, 2006

Re: Wine developers?

by Troy Rollo

On Friday 23 June 2006 22:50, Mike Hearn wrote: > A very specific legal interpretation that would require the > company behind Thinstall to want to hurt the Wine project .... be careful, > none of us are lawyers here. That's not entirely true. -- Troy Rollo - wine(a)troy.rollo.name

June 25, 2006

Re: SecuRom investigation

by n0dalus

On 6/25/06, Martin Owen <doctormo(a)gmail.com> wrote: > > which if we could get our hands on the source code would at least give us > more information about how this security works and may even allow us to work > around it perminently with in wine. > One easy way of improving compatibility with copy protections without any source code is to make all drives visible to wine applications appear to be SCSI. The main copy protections don't "work" on SCSI drives so the games are allowed to run unchecked. I'm not sure how hard this would be to setup in wine at the moment though. n0dalus.

June 25, 2006

Re: [AppDb] Comments handling cleanup

by Jonathan Ernst

Le dimanche 25 juin 2006 à 10:59 -0600, Tony Lambregts a écrit : [...] > I am more in favor of this approach than using makeSafe() and do the same thing > as Chris's query_parameters() patch. However I am hard pressed to say whether > this method is really better or safer than query_parameters(). > > I have had real issues with makeSafe() the primary one is that IMO the place to > make sure that we are safe from sql injection is where we create the SQL. The > makeSafe() did not do that. > > The changes to that are only cosmetic and make it harder to see the actual > changes. Formating changes should be in a separate patch. > > Also I would really appreciate a "Files Changed:" section that lists the files > changed/added/removed by this patch. I find that it really helps in reviewing > patches. > > Please resubmit with the formating changes in a separate patch. Thanks for your comments. I was aware that my changes weren't really atomic but I sent the patch to get some comments. If everyone agrees with the approach I'll make separate patches tomorrow. Thanks

June 25, 2006

Re: SecuRom investigation

by Mike Hearn

On Sun, 25 Jun 2006 02:34:24 +0100, Martin Owen wrote: > which if we could get our hands on the source code would at least give > us more information about how this security works and may even allow us > to work around it perminently with in wine. The goal is not to work around it, the goal is to make it work! That is, it should not be any easier to copy games by using Wine. As for getting the SR code itself, no can do. We just need to solve this one the hard way and that means just bashing away at the bugs one at a time. All copy protection uses kernel drivers these days so the NTOSKRNL work is required first.

June 25, 2006

Re: Patch: winex11drv create_cursor Loading 32bit rgba cursor image

by Mike Hearn

Heya Pauli, A review is enclosed. To submit a patch please send it to the wine-patches list, where it'll enter Alexandres queue for review. Firstly I'm kind of surprised this doesn't require the XCursor extension. Handling the case of a solid alpha channel by converting it to a mask is fine but you should note with a FIXME cases where it doesn't work properly. > @@ -422,16 +422,18 @@ static Cursor create_cursor( Display *di > } > else > { > - int rbits, gbits, bbits, red, green, blue; > + int rbits, gbits, bbits,abits, red, green, blue, alpha; abits doesn't seem to be used anywhere (apart from the assignment in the 32bit case). > + alpha = 255; // This is vissible? Yes, 255 is visible/max opacity (note spelling). Use C /* */ style comments, we require this. > + if (ptr->bBitsPerPixel == 32) > + { > + theChar = theImage[byteIndex++]; > + alpha = theChar; > + } > break; I'm not sure even why "theChar" exists, apart from in one usage it seems to be worthless. You could just directly assign it here (with the right cast). Really this whole function could use a bit of spring cleaning. > + > + //bg or fg? > + // If there is alpha for fg color then we should calculate avarange > + // alpha and use that to create transperancy for cursor (at least my gnome > + // has transperancy with xorg 7) I'm afraid it's really not clear what you're doing here. Standard X cursors cannot have an alpha channel, support for this was added with the XRender/XCursor extensions which you don't seem to be using. Here you seem to have decided that any value larger than 8 for the alpha channel is solid and anything less is transparent. This makes no sense. A much simpler way is to just compare against zero and 255, like this: if (alpha == 0) { /* Transparent (note spelling) */ pAndBits[xorIndex] |= bitShifted; } else if (alpha < 255) { static BOOL warned = FALSE; if (!warned) { FIXME("non-simple alpha channels are currently unsupported, expect visual glitches!\n"); warned = TRUE; } /* assume we should display partly transparent pixels anyway */ } not this ... > + if (alpha < threshold>>3) > + { > + // Transperant > + pAndBits[xorIndex] |= bitShifted; > + } The other thing that confuses me about this code is that on first inspection it appears to be backwards. Surely if the pixel needs to be transparent then that bit of pAndBits should be zero, which is what it's initialized to, and it should be set if the pixel is opaque? I expect I'm confused, after all you tested this. But I'd like somebody to explain it .... > + // What is going on? > +/* TRACE("Cursor file pixel, No dead code please. If you found this helpful that's great but don't submit code that is commented out. > + if (!pixmapBits > + && !pixmapMask) > { > XFreePixmap( display, pixmapAll ); > XFreeGC( display, gc ); This isn't strictly correct, as only the last could have failed. Something like if (!pixmapBits || !pixmapMask) { if (pixmapBits) XFreePixmap( display, pixmapAll ); etc ... } would be better. > @@ -556,6 +592,15 @@ static Cursor create_cursor( Display *di > /* Put the image */ > XCopyArea( display, pixmapBits, pixmapAll, gc, > 0, 0, xmax, ymax, 0, ptr->nHeight ); > + > + if (ptr->bBitsPerPixel == 32) + { + > // Keep original transperancy if not rgba cursor + > XCopyArea( display, pixmapMask, pixmapAll, gc, + > 0, 0, xmax, ymax, 0, 0); + } > + The comment here is a bit backwards, it implies it's on a non-alpha codepath when the reverse is true. It'd be better as /* Apply the alpha channel */ or something like that. Final note - you are using a separate mask when we already have one in this code. Why not merge your new mask with the existing one? thanks -mike

June 25, 2006

Re: An RPM of wine with safedisc support

by Mike Hearn

On Sat, 24 Jun 2006 15:22:42 +0100, ivanleo wrote: > Safedisc 2 and above are not supported. Is the reason why not on the Wiki? Would be cool to get it there if not. > To run a safedisc protected game, you must run winecfg and > set the drive type of your cdrom drive to "cdrom" in the drives tab. > Yes, this sounds dumb, but it has to be done. It sounds dumb because it is dumb, it implies a big failure in our auto detection code. I think the drive autodetect code I wrote was never enabled was it, as we now use HAL anyway, but it seems we really need to fix this bug ... thanks -mike

June 25, 2006

Re: Ole-BSTR-Concat broken?

by Mike Hearn

On Sat, 24 Jun 2006 11:47:51 +0200, Olaf Schmidt wrote: > Do you handle the OLE-BSTR-stuff directly in wine, or do you > give that jobs to other Libs in the system (the 0.9.12er winelib- > update has also changed some dependencies (Gtk, libc* and > others)? BSTR (Basic String) is a Microsoft specific construct, it has no equivalent on other platforms, so it's handled internally. thanks -mike

June 25, 2006

Re: indented relay traces

by Mike Hearn

On Sat, 24 Jun 2006 02:15:09 -0700, James Hawkins wrote: > Ah, didn't know about that tool. You learn something new every day. Perhaps it should be mentioned in the developer quickstart. thanks -mike

June 25, 2006

Re: [AppDb] Comments handling cleanup

by Tony Lambregts

Jonathan Ernst wrote: > Hi, > > I followed your discussions about sql/html injection and had a look at > what you are doing in order to fix the situation and I don't want to > offense anybody but I feel that what you are trying to do is a bit hacky > ;-). > > Here is a cleaned up version of the comments handling that better fixes > (imho) all problems related to sql and html injection. > > The code I made is not dependent on magic_quote value and should'nt > allow anybody to mess with the values read or written from/to the > database. Also it doesn't require to make copies of every single string the user > passes to us. > > It is based on php manual's best practices for avoiding injection. > > I'd be very glad if we'll use such mechanism for the rest of the > queries and get rid of compile_whatever, makeSafe & co. > > This patch follows the Coding Standards we set up a little bit more as > well. > > Awaiting your comments. I am more in favor of this approach than using makeSafe() and do the same thing as Chris's query_parameters() patch. However I am hard pressed to say whether this method is really better or safer than query_parameters(). I have had real issues with makeSafe() the primary one is that IMO the place to make sure that we are safe from sql injection is where we create the SQL. The makeSafe() did not do that. The changes to that are only cosmetic and make it harder to see the actual changes. Formating changes should be in a separate patch. Also I would really appreciate a "Files Changed:" section that lists the files changed/added/removed by this patch. I find that it really helps in reviewing patches. Please resubmit with the formating changes in a separate patch. -- Tony Lambregts

June 25, 2006

Re: [AppDb] stuff thats broken

by Tony Lambregts

Chris Morgan wrote: >>> 5.) If you add a hyperlink in a note it is broken with makesafe() >>> turned on. >> I'm not sure this has anything to do with makeSafe(). Right now makeSafe() >> is effectively disabled and I've carefully reviewed the changes to >> addAppNode.php but I can't see how any of the changes broke notes. >> >> On my local machine after previewing a note I'm seeing any url I put in >> prefixed by the current page url. On the appdb when previewing everything >> appears to be working correctly. >> >> I think this may be a legit bug with the injection changes and I'll fix it >> before committing. >> > > Just tested again here with the injection changes and hypertext(tested with a > url) in notes looks good to me. > > Chris > > > let me clarify that this is with makeSafe() turned on. IE male safe defined as this: function makeSafe($var) { $var = trim(addslashes($var)); return $var; } This has nothing to do with magic_quotes. I have applied your magic_quotes_warning.patch and do not get the warning message that I should if get_magic_quotes_gpc() was on. I have this in my /etc/php.ini ; Magic quotes for incoming GET/POST/Cookie data. magic_quotes_gpc = Off The reason I have makeSafe() "turned on" is that I assume that eventually we will make it use addslashes() in it and I wanted to find any issues that doing so would produce in the course of testing. What I have found is that input fields that use xinha have this issue when makeSafe() uses addslashes(). Does that help... PS: I am still not finished testing the select/update/delete patch -- Tony lambregts

June 25, 2006

Re: Recording with "Lexia" (mci "open new")

by Eric Pouech

> I'm having a simular problem with G711Tool.exe > > 0009:trace:mci:mciSendStringW (L"open waveaudio alias snd", > 0x40371358, 512, (nil)) the same fix should do it... don't forget to apply (if you didn't do it yet) the latest changes to wine.inf so that mciwave.dll is you mci driver (and no longer mciwave.drv) A+

June 25, 2006

Re: Recording with "Lexia" (mci "open new")

by Eric Pouech

Damjan Jovanovic wrote: >>3) Most importantly: Why is it necessary to add >>CAPTURE to system.ini? >> Perhaps the "alias" argument isn't correctly >>handled? >> >> > >Yes wine's MCI is broken: the parsing of some >formats >of MCI commands is incorrect, one of the examples I >clearly remember being: > >"open G: type cdaudio alias cd" >and wine complains that it could not load driver for >type G: (whereas the driver type is cdaudio, no?) >and you have to add "G:=mcicda.dll" in system.ini > > > just sent a fix for this one A+

June 25, 2006

Re: How do I get the unix filename for a wine handle?

by Christoph Probst

Hi. Eric Pouech wrote: > just for sake of completeness: how about enhancing ClamAV so that it > takes a fd (instead of a filename) as its input ? It looks like as if fd are already supported somehow. Need to have a closer look at that ... But I found an even better alternative: ClamAV supports a STREAM command which allows a client to send a string to the scanner. This allows to scan a string even before it is written to disk. Thanks for your answer! Chris -- Kontakt-Details: http://www.christoph-probst.com/kontakt/ PGP-FP: B171 7EA4 988C DD90 1601 D21C 5279 2FAF 9978 AF86

June 25, 2006

Re: riched20: WM_SETFONT support

by Matt Finnicum

On 6/25/06, Krzysztof Foltman <wdev(a)foltman.com> wrote: > I'm definitely for using your conformance test code. And WM_GETFONT. Not > sure about the rest - the code in my patch is based on implementation of > EM_SETCHARFORMAT, so it's basically as good as that one. Actually, GETFONT was a mistake - it turns out it doesn't exist for rich edit controls. > > 1) You need to rewrap the paragraphs, since line lengths change. > > Yes, but SetCharFormat already does it internally (run.c:624), so > there's no need to explicitly call it again. I had some issues with rewrapping not happening when I was testing with mine, but it appears to be working fine on yours. A long line, after being SETFONT'd to a larger font, would have overlapping text. > > 2) Your code ignores lParam (which is supposed to specify if it > > repaints immediately) Ahh. Missed that somehow. > > 3) You need to invalidate the entire control - otherwise just the are > > the text takes up will be redrawn - if the text shrinks, then the area > > outside it's new size won't be blanked / you'll see fragments of the > > old text. > > I'm not sure how the original control is behaving - going with > InvalidateRect, no matter if it's necessary or not, won't harm. However, > it should be done in ME_SetCharFormat, so that both EM_SETCHARFORMAT and > WM_SETFONT (and other functions) will use it. Sounds good - I didn't test how SetCharFormat was behaving, but you're right - it's probably needed by both. The redraw glitch can be pretty obvious (just change any chunk of text to a smaller font) > > 4) Your code (I believe) will reset the scroll position - I store it > > ahead of time and then restore it. > > And that is actually wrong - keep in mind that scroll position is > expressed in pixels, and the new font may be sometimes much smaller than > the old one. > > I think the scroll position update code belongs to ME_SetCharFormat, and > should be added there in a separate patch. Compare WM_SETFONT to > EM_SETCHARFORMAT and you'll know why. Sounds good to me. When it is implemented, I'll add a conformance test for the scrolling behaviour, too. > The plan is: if my original patch gets accepted, then you can extract > the testing code and WM_GETFONT handler which is missing in my > implementation and send it as a separate patch. This will avoid the mess > related to combining those two patches into one. MMkay. Will do! > When this stuff is finished, I'd suggest you try implementing the > ES_PASSWORD-related stuff, combined with WM_SETFONT handling it would > let us fix a couple Bugzilla reports :) It should be pretty > straightforward and independent of other parts of the code. And I'm not > going to touch that stuff now. Most Excellent. ES_PASSWORD (actually I wrote about a third of it already) is only supposed to affect single-line controls, but I could always just implement it for all rich edit controls, and that'll right itself when single-line is implemented. That would probably make more sense, as I doubt anything relies on the "Does nothing if it's not a single-line control" behaviour. Actually, after implementing that there's a bunch of styles I could add somewhat easily, like WS_SETFONT. Thanks for the help / I'll let ya know what I'm working on. --Matt

June 25, 2006

[TOOLS] Re: Gecko installer

by Jonathan Ernst

Le samedi 24 juin 2006 à 20:06 +0200, Jacek Caban a écrit : > Hi, > [...] > - We need php script redirecting to SourceForge similar to one that > forwards to Mozilla ActiveX control. Here you are.

June 25, 2006

Re: Patch: winex11drv create_cursor Loading 32bit rgba cursor image

by Pauli

I looked the code bit more. Then I thought I hate if and removed some of them. Now I think code is cleaner but it has minor performance hit when cursor don't have alpha. But create_cursor won't be called every frame.

June 25, 2006

Re: wined3d: declaration/FVF conversion test

by Jason Green

On 6/19/06, Ivan Gyurdiev <ivg231(a)gmail.com> wrote: > Does the windows box support shader model 3.0 ? The reason I keep asking > about this is that MSDN talks about a ' pre-3.0' driver and a 3.0 driver, > and I wanted to see if behavior was different. Just tested on my primary machine which has an NVidia 6600GT (Shader Model 3.0) in Windows XP, and got this: vertexdeclaration: 441 tests executed, 0 marked as todo, 0 failures. So, looks like we are good to go.

June 25, 2006

Re: riched20: WM_SETFONT support

by Krzysztof Foltman

Matt Finnicum wrote: > I was just about to send in the same patch when you did. I'll have to > keep you updated on what I'm working on in the future. That'd be nice - as I mostly stopped doing riched20 stuff for quite some time, and only got back to it a few days ago. > I've attached the patch I was going to send - I suggest we use some > parts of each (either I can merge them into a patch or you can, your > call). I'm definitely for using your conformance test code. And WM_GETFONT. Not sure about the rest - the code in my patch is based on implementation of EM_SETCHARFORMAT, so it's basically as good as that one. > 1) You need to rewrap the paragraphs, since line lengths change. Yes, but SetCharFormat already does it internally (run.c:624), so there's no need to explicitly call it again. > 2) Your code ignores lParam (which is supposed to specify if it > repaints immediately) It does not, it assigns it to bRepaint and then uses in a condition. So it's a correct solution. > 3) You need to invalidate the entire control - otherwise just the are > the text takes up will be redrawn - if the text shrinks, then the area > outside it's new size won't be blanked / you'll see fragments of the > old text. I'm not sure how the original control is behaving - going with InvalidateRect, no matter if it's necessary or not, won't harm. However, it should be done in ME_SetCharFormat, so that both EM_SETCHARFORMAT and WM_SETFONT (and other functions) will use it. > 4) Your code (I believe) will reset the scroll position - I store it > ahead of time and then restore it. And that is actually wrong - keep in mind that scroll position is expressed in pixels, and the new font may be sometimes much smaller than the old one. I think the scroll position update code belongs to ME_SetCharFormat, and should be added there in a separate patch. Compare WM_SETFONT to EM_SETCHARFORMAT and you'll know why. The plan is: if my original patch gets accepted, then you can extract the testing code and WM_GETFONT handler which is missing in my implementation and send it as a separate patch. This will avoid the mess related to combining those two patches into one. When this stuff is finished, I'd suggest you try implementing the ES_PASSWORD-related stuff, combined with WM_SETFONT handling it would let us fix a couple Bugzilla reports :) It should be pretty straightforward and independent of other parts of the code. And I'm not going to touch that stuff now. Krzysztof

June 25, 2006

Re: wine-1.0 goal: fix all reported msi and ole bugs?

by Vincent Povirk

On 6/24/06, Ivan Leo Puoti <ivanleo(a)gmail.com> wrote: > Are you implying 1.0 should be released with any known bugs at all? I thought only M$ did that sort > of stuff. I've never heard of any open source project waiting for all known bugs to be closed before going out of beta. There's always something more to be improved or fixed. 1.0 doesn't mean "perfect"; as I understand it, it means "good enough for normal people to use". That means not having lots of regressions all the time, and not needing extensive configuration before it will run anything properly. But then I'm not really a developer. -- Vincent Povirk

June 25, 2006

Re: SecuRom investigation

by Jonathan Wilson

I dont know for sure but I believe modern versions of SecuRom use kernel mode code/drivers. If so, the work done on NTOSKRNL for SafeDisk might work.

June 25, 2006

SecuRom investigation

by Martin Owen

Hello all, I have been investigating getting rid of a terminal anoyancy for wine game users, SecuRom security. I don't believe it is good enough to crack the games as this makes us out to look like criminals rather than trying to get legitimate games to work. I have sent a query to sony, asking if they will help with providing information, but I'm not hopeful of a positive response. I found an interesting tool here: http://www.gameburnworld.com/securompatches.htm which if we could get our hands on the source code would at least give us more information about how this security works and may even allow us to work around it perminently with in wine. Best Regards, Martin Owens

June 25, 2006

Re: riched20: WM_SETFONT support

by Matt Finnicum

I was just about to send in the same patch when you did. I'll have to keep you updated on what I'm working on in the future. (You can see what I'm planning to do for my SoC project at http://wiki.winehq.org/MatthewFinnicum) I've attached the patch I was going to send - I suggest we use some parts of each (either I can merge them into a patch or you can, your call). Our code is actually quite similar - though your "ME_CharFormatFromLogFont" is definitely superior to mine (I just did it all in function / was slightly less awesome). Aside from my conformance tests (Which your patch passed just fine), my patch covers some extra things: 1) You need to rewrap the paragraphs, since line lengths change. 2) Your code ignores lParam (which is supposed to specify if it repaints immediately) 3) You need to invalidate the entire control - otherwise just the are the text takes up will be redrawn - if the text shrinks, then the area outside it's new size won't be blanked / you'll see fragments of the old text. 4) Your code (I believe) will reset the scroll position - I store it ahead of time and then restore it. Thanks, --Matt On 6/24/06, Krzysztof Foltman <wdev(a)foltman.com> wrote: > ChangeLog: > * WM_SETFONT implementation - sets the default font and the font for > the whole pre-existing text > > Seems to fix bug 4563 and half of the bug 3917. > > Krzysztof > > > diff --git a/dlls/riched20/editor.c b/dlls/riched20/editor.c > old mode 100644 > new mode 100755 > index 50527ca..61ecc72 > --- a/dlls/riched20/editor.c > +++ b/dlls/riched20/editor.c > @@ -92,7 +92,7 @@ > - EM_SELECTIONTYPE > - EM_SETBIDIOPTIONS 3.0 > + EM_SETBKGNDCOLOR > - - EM_SETCHARFORMAT (partly done, no ANSI) > + + EM_SETCHARFORMAT (partly done, no ANSI) > - EM_SETEDITSTYLE > + EM_SETEVENTMASK (few notifications supported) > - EM_SETFONTSIZE > @@ -135,7 +135,7 @@ > + WM_GETTEXT (ANSI&Unicode) > + WM_GETTEXTLENGTH (ANSI version sucks) > + WM_PASTE > - - WM_SETFONT > + + WM_SETFONT > + WM_SETTEXT (resets undo stack !) (proper style?) ANSI&Unicode > - WM_STYLECHANGING > - WM_STYLECHANGED (things like read-only flag) > @@ -1442,7 +1442,6 @@ LRESULT WINAPI RichEditANSIWndProc(HWND > UNSUPPORTED_MSG(EM_SETTYPOGRAPHYOPTIONS) > UNSUPPORTED_MSG(EM_SETWORDBREAKPROCEX) > UNSUPPORTED_MSG(EM_SHOWSCROLLBAR) > - UNSUPPORTED_MSG(WM_SETFONT) > UNSUPPORTED_MSG(WM_STYLECHANGING) > UNSUPPORTED_MSG(WM_STYLECHANGED) > /* UNSUPPORTED_MSG(WM_UNICHAR) FIXME missing in Wine headers */ > @@ -1844,6 +1843,27 @@ LRESULT WINAPI RichEditANSIWndProc(HWND > > return 0; > } > + case WM_SETFONT: > + { > + LOGFONTW lf; > + CHARFORMAT2W fmt; > + HDC hDC; > + BOOL bRepaint = LOWORD(lParam); > + > + if (!wParam) > + wParam = (WPARAM)GetStockObject(DEFAULT_GUI_FONT); > + GetObjectW((HGDIOBJ)wParam, sizeof(LOGFONTW), &lf); > + hDC = GetDC(hWnd); > + ME_CharFormatFromLogFont(hDC, &lf, &fmt); > + ReleaseDC(hWnd, hDC); > + ME_SetCharFormat(editor, 0, ME_GetTextLength(editor), &fmt); > + ME_SetDefaultCharFormat(editor, &fmt); > + > + ME_CommitUndo(editor); > + if (bRepaint) > + ME_UpdateRepaint(editor); > + return 0; > + } > case WM_SETTEXT: > { > ME_InternalDeleteText(editor, 0, ME_GetTextLength(editor)); > diff --git a/dlls/riched20/editor.h b/dlls/riched20/editor.h > old mode 100644 > new mode 100755 > index 33b7626..a4d4980 > --- a/dlls/riched20/editor.h > +++ b/dlls/riched20/editor.h > @@ -55,6 +55,7 @@ void ME_CopyToCF2W(CHARFORMAT2W *to, CHA > CHARFORMAT2W *ME_ToCFAny(CHARFORMAT2W *to, CHARFORMAT2W *from); > void ME_CopyToCFAny(CHARFORMAT2W *to, CHARFORMAT2W *from); > void ME_CopyCharFormat(CHARFORMAT2W *pDest, CHARFORMAT2W *pSrc); /* only works with 2W structs */ > +void ME_CharFormatFromLogFont(HDC hDC, LOGFONTW *lf, CHARFORMAT2W *fmt); /* ditto */ > > /* list.c */ > void ME_InsertBefore(ME_DisplayItem *diWhere, ME_DisplayItem *diWhat); > diff --git a/dlls/riched20/style.c b/dlls/riched20/style.c > old mode 100644 > new mode 100755 > index f420f09..6104a0c > --- a/dlls/riched20/style.c > +++ b/dlls/riched20/style.c > @@ -286,6 +286,25 @@ ME_LogFontFromStyle(HDC hDC, LOGFONTW *l > lf->lfCharSet = s->fmt.bCharSet; > } > > +void ME_CharFormatFromLogFont(HDC hDC, LOGFONTW *lf, CHARFORMAT2W *fmt) > +{ > + int rx, ry; > + > + ME_InitCharFormat2W(fmt); > + rx = GetDeviceCaps(hDC, LOGPIXELSX); > + ry = GetDeviceCaps(hDC, LOGPIXELSY); > + lstrcpyW(fmt->szFaceName, lf->lfFaceName); > + fmt->dwEffects = 0; > + fmt->dwMask = CFM_WEIGHT|CFM_BOLD|CFM_ITALIC|CFM_UNDERLINE|CFM_STRIKEOUT|CFM_SIZE|CFM_FACE|CFM_CHARSET; > + fmt->wWeight = lf->lfWeight; > + fmt->yHeight = -lf->lfHeight*1440/ry; > + if (lf->lfWeight>400) fmt->dwEffects |= CFM_BOLD; > + if (lf->lfItalic) fmt->dwEffects |= CFM_ITALIC; > + if (lf->lfUnderline) fmt->dwEffects |= CFM_UNDERLINE; > + if (lf->lfStrikeOut) fmt->dwEffects |= CFM_STRIKEOUT; > + fmt->bPitchAndFamily = lf->lfPitchAndFamily; > + fmt->bCharSet = lf->lfCharSet; > +} > > BOOL ME_IsFontEqual(LOGFONTW *p1, LOGFONTW *p2) > { > > > > > >

June 25, 2006

Re: Autocad 2004 STATUS_INVALID_LDT_OFFSET

by Jaap Stolk

On 6/25/06, Ivan Leo Puoti <ivanleo(a)gmail.com> wrote: > Jaap Stolk wrote: > > from what I could google, code=c0000096 means that a privileged > > instruction was found. > > This is to be expected. DispatchDeviceControl is the function in the driver which handles the IO for > the kernel (in this case our fake kernel). If it's a driver it obviously must need to do something > that has to be done in ring 0, such as running a privileged instruction. For safedisc, we have some > hardware emulation which will run the privileged instructions it needs. What you'll probably have to > do is find the instructions the particular driver needs (you can find it in > c:\windows\system32\drivers) and emulate it somehow, which may or may not be possible. > > Ivan. Thanks Ivan. I did some IRC with Vitaliy, and he came to the conclusion that the outport() code was probably the result of messed up code/stack, since the program doesn't use any hardware or locks. I will look into it tomorrow. Jaap.

June 25, 2006

Re: An RPM of wine with safedisc support

by Vijay Kiran Kamuju

Hi, I am just wanting to keep the patch as a diff somewhere like the ddraw patch maintained by Stephan Dosinger. Thanks, Vijay On 6/24/06, ivanleo(a)gmail.com <ivanleo(a)gmail.com> wrote: > On 24/06/06, Vijay Kiran Kamuju <infyquest(a)gmail.com> wrote: > > Can you tell us what are all the specific pathces went in to this rpm > > with releated to safedisc? > > > > I know that Vitaly's ntoskernel patch is isn, what else. > All the safedisc related code developed in the last year or so, I can > send you a full diff against winehq if you want. > > Ivan. >

June 24, 2006

Re: Autocad 2004 STATUS_INVALID_LDT_OFFSET

by Ivan Leo Puoti

Jaap Stolk wrote: > from what I could google, code=c0000096 means that a privileged > instruction was found. This is to be expected. DispatchDeviceControl is the function in the driver which handles the IO for the kernel (in this case our fake kernel). If it's a driver it obviously must need to do something that has to be done in ring 0, such as running a privileged instruction. For safedisc, we have some hardware emulation which will run the privileged instructions it needs. What you'll probably have to do is find the instructions the particular driver needs (you can find it in c:\windows\system32\drivers) and emulate it somehow, which may or may not be possible. Ivan.

June 24, 2006

Re: Autocad 2004 STATUS_INVALID_LDT_OFFSET

by Vitaliy Margolen

Saturday, June 24, 2006, 4:33:03 PM, Jaap Stolk wrote: > On 6/25/06, Vitaliy Margolen <wine-devel(a)kievinfo.com> wrote: >> It's pretty hard to see what you doing from this pastes. Do you have time to >> join one of the Wine's IRC channels? > never used that before. I seem to have x-chat. where do i find the > correct channel ? http://www.winehq.org/site/irc

June 24, 2006

Patch: winex11drv create_cursor Loading 32bit rgba cursor image

by Pauli

fix for bug 5486 Patch makes it possible to load and display 32bit rgba cursor. I have made tested the code with Football Manager 2006. The ealier version of patch was confimed loading 32bit cursors from Guild Wars. I would like to hear comments how to improve the code. Is there chance that this could be included to wine? (without huge modifications) The code is a hack and not perfect solution. But hack is better than invisible cursor.

June 24, 2006

Re: Autocad 2004 STATUS_INVALID_LDT_OFFSET

by Jaap Stolk

On 6/25/06, Vitaliy Margolen <wine-devel(a)kievinfo.com> wrote: > It's pretty hard to see what you doing from this pastes. Do you have time to > join one of the Wine's IRC channels? never used that before. I seem to have x-chat. where do i find the correct channel ?

June 24, 2006

Re: Autocad 2004 STATUS_INVALID_LDT_OFFSET

by Vitaliy Margolen

Saturday, June 24, 2006, 3:18:03 PM, Jaap Stolk wrote: > On 6/24/06, Jaap Stolk <jwstolk(a)gmail.com> wrote: >> So the problem is somewhere in load_driver( ). Is there a way to >> narrow it down a bit more? > using objdump I got a bit closer. > (note that the exact error address my not be consistent, because I > added some more trace functions to the code while testing) > objdump ntoskrnl.exe.so -S > tmp.txt ImagePath ->> 0x1d7fa > 0x1d7fa + 0x4fc = 0x1DCF6 0x1DCF6 ->> static NTSTATUS driver_control(void) > also, combining the load address and the error location ends up somewhere there: > 0x4062de4e - 0x40610000 = 0x1DE4E > 1de21: 83 ec 0c sub $0xc,%esp > 1de24: 52 push %edx > 1de25: 8d 83 9c 02 ff ff lea 0xffff029c(%ebx),%eax > 1de2b: 50 push %eax > 1de2c: 8d 83 a4 01 ff ff lea 0xffff01a4(%ebx),%eax > 1de32: 50 push %eax > 1de33: 8d 83 c8 fe ff ff lea 0xfffffec8(%ebx),%eax > 1de39: 50 push %eax > 1de3a: 6a 01 push $0x1 > 1de3c: e8 c7 5f fe ff call 3e08 <_init+0x2c> > 1de41: 83 c4 20 add $0x20,%esp > 1de44: e9 3d ff ff ff jmp 1dd86 <driver_control+0x1ca> > 1de49: 8b 45 b0 mov 0xffffffb0(%ebp),%eax > 1de4c: 85 c0 test %eax,%eax >>>1de4e: 0f 84 1a ff ff ff je 1dd6e <driver_control+0x1b2> > 1de54: 83 ec 08 sub $0x8,%esp > 1de57: 56 push %esi > 1de58: ff b5 50 fe ff ff pushl 0xfffffe50(%ebp) > 1de5e: ff d0 call *%eax > (it's called form here:) > WINE_ERR("DispatchDeviceControl returned %lx\n", irp.IoStatus.Status); > 1dd75: 8a 83 c8 fe ff ff mov 0xfffffec8(%ebx),%al > 1dd7b: 83 e0 02 and $0x2,%eax > 1dd7e: 84 c0 test %al,%al > 1dd80: 0f 85 9b 00 00 00 jne 1de21 <driver_control+0x265> > looking at it form an other angle: I get two "raise exception". > The first one is just after user32.SetDeskWallPaper( ). I added the > wallpaper patch, but it's still happening. as it happens very early in > the log, it does not seem to be a problem. > I think the second one is causing the debugger to start: > 000f:Ret ntoskrnl.exe.IoCreateSymbolicLink() retval=00000000 ret=407503ae > 000f:trace:ntoskrnl:driver_control returned from call with status 00000000 > 000f:trace:seh:raise_exception code=c0000096 flags=0 addr=0x4062e04c > 000f:trace:seh:raise_exception eax=4062e04c ebx=4063d964 ecx=400d03e8 > edx=0000004b esi=4074fd80 edi=00000000 > 000f:trace:seh:raise_exception ebp=4074fe08 esp=4074fc2c cs=0023 > ds=002b es=002b fs=003b gs=0033 flags=00010212 > 000f:trace:seh:call_stack_handlers calling handler at 0x4033ce80 > code=c0000096 flags=0 > <snip 13 lines> > 000f:Call ntdll.RtlInitAnsiString(4074f5a0,4039d8ed "winedos.dll") ret=403500a0 > <snip 22 pages> > 000f:trace:module:load_builtin_callback loaded winedos.dll 0x4020d888 0x40760000 > from what I could google, code=c0000096 means that a privileged > instruction was found. > But I don't seen anything exotic at 0x1de4e. > I think this is related: > http://bugs.winehq.com/show_bug.cgi?id=3438 > but winedos.dll is started after the problem, so i don't think it's causing it. > I'm still not sure how to track the error location to the correct line > of source code, I recompiled ntoskrnl with -ggdb, but it already had > -g, so it should contain the source line numbers right? > How does wine normally handle privileged instruction ? I'm sure the > safedisc code uses every privileged instruction it can, and i would > not be surprised if it was throwing exceptions as a normal code path, > and i suspect some of it's drivers are supposed to run in ring 0 mode. It's pretty hard to see what you doing from this pastes. Do you have time to join one of the Wine's IRC channels? What I see there the driver tries to do soemthing that it thinks could be done running inside the kernel. But we don't run inside the kernel so those instructions do raise exception. Depending on what it's trying to do we should be able to emulate what it needs (if you applied all of my patch, especially the part to kernel/except.c Vitaliy

June 24, 2006

Re: Autocad 2004 STATUS_INVALID_LDT_OFFSET

by Jaap Stolk

On 6/24/06, Jaap Stolk <jwstolk(a)gmail.com> wrote: > So the problem is somewhere in load_driver( ). Is there a way to > narrow it down a bit more? using objdump I got a bit closer. (note that the exact error address my not be consistent, because I added some more trace functions to the code while testing) objdump ntoskrnl.exe.so -S > tmp.txt ImagePath -> 0x1d7fa 0x1d7fa + 0x4fc = 0x1DCF6 0x1DCF6 -> static NTSTATUS driver_control(void) also, combining the load address and the error location ends up somewhere there: 0x4062de4e - 0x40610000 = 0x1DE4E 1de21: 83 ec 0c sub $0xc,%esp 1de24: 52 push %edx 1de25: 8d 83 9c 02 ff ff lea 0xffff029c(%ebx),%eax 1de2b: 50 push %eax 1de2c: 8d 83 a4 01 ff ff lea 0xffff01a4(%ebx),%eax 1de32: 50 push %eax 1de33: 8d 83 c8 fe ff ff lea 0xfffffec8(%ebx),%eax 1de39: 50 push %eax 1de3a: 6a 01 push $0x1 1de3c: e8 c7 5f fe ff call 3e08 <_init+0x2c> 1de41: 83 c4 20 add $0x20,%esp 1de44: e9 3d ff ff ff jmp 1dd86 <driver_control+0x1ca> 1de49: 8b 45 b0 mov 0xffffffb0(%ebp),%eax 1de4c: 85 c0 test %eax,%eax >>1de4e: 0f 84 1a ff ff ff je 1dd6e <driver_control+0x1b2> 1de54: 83 ec 08 sub $0x8,%esp 1de57: 56 push %esi 1de58: ff b5 50 fe ff ff pushl 0xfffffe50(%ebp) 1de5e: ff d0 call *%eax (it's called form here:) WINE_ERR("DispatchDeviceControl returned %lx\n", irp.IoStatus.Status); 1dd75: 8a 83 c8 fe ff ff mov 0xfffffec8(%ebx),%al 1dd7b: 83 e0 02 and $0x2,%eax 1dd7e: 84 c0 test %al,%al 1dd80: 0f 85 9b 00 00 00 jne 1de21 <driver_control+0x265> looking at it form an other angle: I get two "raise exception". The first one is just after user32.SetDeskWallPaper( ). I added the wallpaper patch, but it's still happening. as it happens very early in the log, it does not seem to be a problem. I think the second one is causing the debugger to start: 000f:Ret ntoskrnl.exe.IoCreateSymbolicLink() retval=00000000 ret=407503ae 000f:trace:ntoskrnl:driver_control returned from call with status 00000000 000f:trace:seh:raise_exception code=c0000096 flags=0 addr=0x4062e04c 000f:trace:seh:raise_exception eax=4062e04c ebx=4063d964 ecx=400d03e8 edx=0000004b esi=4074fd80 edi=00000000 000f:trace:seh:raise_exception ebp=4074fe08 esp=4074fc2c cs=0023 ds=002b es=002b fs=003b gs=0033 flags=00010212 000f:trace:seh:call_stack_handlers calling handler at 0x4033ce80 code=c0000096 flags=0 <snip 13 lines> 000f:Call ntdll.RtlInitAnsiString(4074f5a0,4039d8ed "winedos.dll") ret=403500a0 <snip 22 pages> 000f:trace:module:load_builtin_callback loaded winedos.dll 0x4020d888 0x40760000 from what I could google, code=c0000096 means that a privileged instruction was found. But I don't seen anything exotic at 0x1de4e. I think this is related: http://bugs.winehq.com/show_bug.cgi?id=3438 but winedos.dll is started after the problem, so i don't think it's causing it. I'm still not sure how to track the error location to the correct line of source code, I recompiled ntoskrnl with -ggdb, but it already had -g, so it should contain the source line numbers right? How does wine normally handle privileged instruction ? I'm sure the safedisc code uses every privileged instruction it can, and i would not be surprised if it was throwing exceptions as a normal code path, and i suspect some of it's drivers are supposed to run in ring 0 mode.

June 24, 2006

Re: Gecko installer

by Dimi Paun

On Sat, 2006-06-24 at 20:06 +0200, Jacek Caban wrote: > - Uploading wine_gecko.cab to SourceForge. I don't have an access to > do so. Could someone that can do it contact me privately? Or perhaps I > could get access so that in the future, when it needs to be updated I > could do it without involving others? Yes, I can give you rights to upload. Give me a few minutes... -- Dimi Paun <dimi(a)lattica.com> Lattica, Inc.

June 24, 2006

Re: wine-1.0 goal: fix all reported msi and ole bugs?

by Ivan Leo Puoti

Dan Kegel wrote: > I just looked, and there aren't *that* many msi and ole bugs > reported in bugzilla. Are you implying 1.0 should be released with any known bugs at all? I thought only M$ did that sort of stuff. Ivan.

June 24, 2006

Re: An RPM of wine with safedisc support

by ivanleo＠gmail.com

On 24/06/06, Vijay Kiran Kamuju <infyquest(a)gmail.com> wrote: > Can you tell us what are all the specific pathces went in to this rpm > with releated to safedisc? > > I know that Vitaly's ntoskernel patch is isn, what else. All the safedisc related code developed in the last year or so, I can send you a full diff against winehq if you want. Ivan.

June 24, 2006

Re: Gecko installer

by Jacek Caban

Hi, Now that Wine Gecko installer is in the tree, we have to enable it. All we have to do on Wine side is set one registry value, but first we need to have wine_gecko.cab file available to download. It means: - Uploading wine_gecko.cab to SourceForge. I don't have an access to do so. Could someone that can do it contact me privately? Or perhaps I could get access so that in the future, when it needs to be updated I could do it without involving others? - We need php script redirecting to SourceForge similar to one that forwards to Mozilla ActiveX control. Thanks, Jacek

June 24, 2006

Re: An RPM of wine with safedisc support

by Vijay Kiran Kamuju

Can you tell us what are all the specific pathces went in to this rpm with releated to safedisc? I know that Vitaly's ntoskernel patch is isn, what else. Thanks, Vijay On 6/24/06, ivanleo(a)gmail.com <ivanleo(a)gmail.com> wrote: > Safedisc has now been working on wine for almost a year, but it'll > still take some time before we get the code into winehq, so I've > decided to build an RPM of wine with safedisc 1 support so users can > play around with it, you can download it from > http://www003.portalis.it/115/wine-safedisc-0.9.16-mdk.i586.rpm > Safedisc 2 and above are not supported. This is based on wine 0.9.16, > however it isn't an official winehq release, and is provided on an as > is basis. To run a safedisc protected game, you must run winecfg and > set the drive type of your cdrom drive to "cdrom" in the drives tab. > Yes, this sounds dumb, but it has to be done. Then run the game > normally, and as long as you have an original CD (Ok, copies can work > too but that's Macrovision's fault) in the drive, the game will start. > > Ivan. > > >

June 24, 2006

An RPM of wine with safedisc support

by ivanleo＠gmail.com

Safedisc has now been working on wine for almost a year, but it'll still take some time before we get the code into winehq, so I've decided to build an RPM of wine with safedisc 1 support so users can play around with it, you can download it from http://www003.portalis.it/115/wine-safedisc-0.9.16-mdk.i586.rpm Safedisc 2 and above are not supported. This is based on wine 0.9.16, however it isn't an official winehq release, and is provided on an as is basis. To run a safedisc protected game, you must run winecfg and set the drive type of your cdrom drive to "cdrom" in the drives tab. Yes, this sounds dumb, but it has to be done. Then run the game normally, and as long as you have an original CD (Ok, copies can work too but that's Macrovision's fault) in the drive, the game will start. Ivan.

June 24, 2006

Re: How do I get the unix filename for a wine handle?

by Kari Hurtta

Eric Pouech <eric.pouech(a)wanadoo.fr> writes: > Christoph Probst wrote: > > >Hi. > > > >Alexandre Julliard wrote: > > > > > >>You can't do that in general. In Unix a file can have multiple names, > >>or even none at all, there's simply no way to get a filename from a > >>handle. On Linux you can use /proc/self/fd but that's not very > >>portable. > >> > > > > So what do you think about the solution that the wineserver stores > > the filename in the file object when the object is created? > > > > Just like wine_server_handle_to_fd() there would be a > > wine_server_handle_to_filename() that returns the filename. > > > >Any objections to this? > > > > > just for sake of completeness: how about enhancing ClamAV so that it > takes a fd (instead of a filename) as its input ? > A+ It does http://www.clamav.net/doc/0.88.2/html/node24.html | The daemon is fully configurable via the clamd.conf file 5. clamd recognizes the following commands: <...> | STREAM Scan stream: clamd will return a new port number you should connect to and send data to scan. You not need to use following command: | SCAN file/directory Scan file or directory (recursively) with archive support enabled (a full path is required).

June 24, 2006

Re: How do I get the unix filename for a wine handle?

by Eric Pouech

Christoph Probst wrote: >Hi. > >Alexandre Julliard wrote: > > > >>You can't do that in general. In Unix a file can have multiple names, >>or even none at all, there's simply no way to get a filename from a >>handle. On Linux you can use /proc/self/fd but that's not very >>portable. >> >> > >So what do you think about the solution that the wineserver stores the >filename in the file object when the object is created? > >Just like wine_server_handle_to_fd() there would be a >wine_server_handle_to_filename() that returns the filename. > >Any objections to this? > > > just for sake of completeness: how about enhancing ClamAV so that it takes a fd (instead of a filename) as its input ? A+

June 24, 2006

Re: How do I get the unix filename for a wine handle?

by Alexandre Julliard

Christoph Probst <chris-usenet(a)netzpunkt.org> writes: > So what do you think about the solution that the wineserver stores the > filename in the file object when the object is created? > > Just like wine_server_handle_to_fd() there would be a > wine_server_handle_to_filename() that returns the filename. > > Any objections to this? Yes. Like I said, there's no guarantee that the file even has a name, or that it is accessible under the name that was used to open it, so we can't provide such an API. -- Alexandre Julliard julliard(a)winehq.org

June 24, 2006

Re: How do I get the unix filename for a wine handle?

by Christoph Probst

Hi. Alexandre Julliard wrote: > You can't do that in general. In Unix a file can have multiple names, > or even none at all, there's simply no way to get a filename from a > handle. On Linux you can use /proc/self/fd but that's not very > portable. So what do you think about the solution that the wineserver stores the filename in the file object when the object is created? Just like wine_server_handle_to_fd() there would be a wine_server_handle_to_filename() that returns the filename. Any objections to this? Chris -- Kontakt-Details: http://www.christoph-probst.com/kontakt/ PGP-FP: B171 7EA4 988C DD90 1601 D21C 5279 2FAF 9978 AF86

June 24, 2006

Re: Autocad 2004 STATUS_INVALID_LDT_OFFSET

by Jaap Stolk

found another one :-) I reconstructed the debug output mixed with the wine log: ------------------------------------ Unhandled exception: page fault on read access to 0x00000000 in 32-bit code (0x4062de4e). Register dump: CS:0023 SS:002b DS:002b ES:002b FS:003b GS:0033 EIP:4062de4e ESP:4074fc2c EBP:4074fe08 EFLAGS:00010212( - 00 - RIA1) EAX:4062de4c EBX:4063c8c4 ECX:400d03e8 EDX:0000004b ESI:4074fd81 EDI:00000000 Stack dump: 0x4074fc2c: 40622e4a 00000020 4074fd80 00000078 0x4074fc3c: 00000000 400d040f 400d042d 400d07e7 0x4074fc4c: 00000078 400d07e7 4063c940 00000020 0x4074fc5c: 083c7748 00000000 00ff4007 401fcb80 0x4074fc6c: 401fcac0 00000000 00000000 00000000 0x4074fc7c: 00000000 00000000 00000000 00000000 Backtrace: =>1 0x4062de4e ImagePath+0x4fc in ntoskrnl (0x4062de4e) 2 0x406226c9 __wine_spec_exe_entry+0x12d(peb=0x400d0800) [/home/jaap/wine-git/programs/ntoskrnl/../../include/wine/list.h:132] in ntoskrnl (0x406226c9) 3 0x40374eef start_process+0x11b(arg=0x0) [/home/jaap/wine-git/dlls/kernel/process.c:832] in kernel32 (0x40374eef) 4 0x4001c233 wine_switch_to_stack+0x17 in libwine.so.1 (0x4001c233) 0x4062de4e ImagePath+0x4fc in ntoskrnl: imull $0x20642528,0x0(%edi),%edx Modules: Module Address Debug info Name (41 modules) <snip> ELF 0x40605000-4063e000 Stabs ntoskrnl<elf> \-PE 0x40610000-4063e000 \ ntoskrnl ------------------------------------ I'm not very experienced with Linux debug outputs but I assume "ImagePath+0x4fc" is somewhere after "static const WCHAR ImagePath[]" in ntoskrnl.c So the problem is somewhere in load_driver( ). Is there a way to narrow it down a bit more? Jaap.

June 24, 2006

Re: Very large files

by Marcus Meissner

On Fri, Jun 23, 2006 at 01:48:31PM -0700, John Willis wrote: > Howdy, > > I have a large VS 6.0 C++ program and I'm running it under WINE under Redhat Linux. > Everything works fine, but it doesn't handle file sizes greater than 4gb like it can running > directly under Windows XP. > I've tweaked everything that I can find that seems to deal with file sizes. > Is there a problem with WINE itself handling very large file sizes? It always depends on what functions it uses. Not all places in WINE might be working for files larger than 4GB. Can you give us +relay,+file traces of such failures? Ciao, Marcus

June 24, 2006

Re: Ole-BSTR-Concat broken?

by Marcus Meissner

On Sat, Jun 24, 2006 at 11:47:51AM +0200, Olaf Schmidt wrote: > Hi *.*, > > seems, that the Ole-BSTR-Concat doesn't work anymore: > (Debian Sid, from wine 0.9.12 upwards - wine configured, > to use the builtin Ole-Stuff) > > In VB-Applications (wich are using (wide) Ole-BSTRs under > the hood), I can reduce the problem to the following: > > That means, that the current implementation does not use > the BSTR-Len-Descriptor anymore (it has worked in 0.9.10 > or 0.9.11 - not very sure, but think it was 0.9.11 in my last tests) > Instead it now behaves like routines, wich have to concat > Zero-Terminated Strings. > > Nothing changed in my wine-configuration - only apt-getted to > 0.9.12 first -> then to the latest 0.9.16 (ubuntu-deb) -> same > problem. > > Do you handle the OLE-BSTR-stuff directly in wine, or do you > give that jobs to other Libs in the system (the 0.9.12er winelib- > update has also changed some dependencies (Gtk, libc* and > others)? We do this ourselves. The current CVS contains fixes from Juan Lang for this or at least a related problem. Ciao, Marcus

June 24, 2006

Re: Ole-BSTR-Concat broken?

by Neil Skrypuch

On Saturday, June 24, 2006 05:47, Olaf Schmidt wrote: > Hi *.*, > > seems, that the Ole-BSTR-Concat doesn't work anymore: > (Debian Sid, from wine 0.9.12 upwards - wine configured, > to use the builtin Ole-Stuff) > > In VB-Applications (wich are using (wide) Ole-BSTRs under > the hood), I can reduce the problem to the following: > > If I define two Strings with the Content: > > S1, containing a Zero WChar > (BSTR-LenDescriptor=2, ByteSequence 0 0) > > and S2, containing "A|Zero|B" > (BSTR-LenDescriptor=6, ByteSequence 65 0 | 0 0 | 66 0) > > and then do the concat: > SResult = S1 + S2 > > the resulting string contains only an "A" > (BSTR-LenDescriptor=2, ByteSequence 65 0) > > instead of the correct ByteSequence: 0 0 | 65 0 | 0 0 | 66 0 > > That means, that the current implementation does not use > the BSTR-Len-Descriptor anymore (it has worked in 0.9.10 > or 0.9.11 - not very sure, but think it was 0.9.11 in my last tests) > Instead it now behaves like routines, wich have to concat > Zero-Terminated Strings. > > Nothing changed in my wine-configuration - only apt-getted to > 0.9.12 first -> then to the latest 0.9.16 (ubuntu-deb) -> same > problem. > > Do you handle the OLE-BSTR-stuff directly in wine, or do you > give that jobs to other Libs in the system (the 0.9.12er winelib- > update has also changed some dependencies (Gtk, libc* and > others)? > > Olaf Schmidt If you could do a regression test and find the specific patch that broke things, that would be very helpful. See section 6.1 of http://wiki.winehq.org/GitWine for regression testing. - Neil

June 24, 2006

Ole-BSTR-Concat broken?

by Olaf Schmidt

Hi *.*, seems, that the Ole-BSTR-Concat doesn't work anymore: (Debian Sid, from wine 0.9.12 upwards - wine configured, to use the builtin Ole-Stuff) In VB-Applications (wich are using (wide) Ole-BSTRs under the hood), I can reduce the problem to the following: If I define two Strings with the Content: S1, containing a Zero WChar (BSTR-LenDescriptor=2, ByteSequence 0 0) and S2, containing "A|Zero|B" (BSTR-LenDescriptor=6, ByteSequence 65 0 | 0 0 | 66 0) and then do the concat: SResult = S1 + S2 the resulting string contains only an "A" (BSTR-LenDescriptor=2, ByteSequence 65 0) instead of the correct ByteSequence: 0 0 | 65 0 | 0 0 | 66 0 That means, that the current implementation does not use the BSTR-Len-Descriptor anymore (it has worked in 0.9.10 or 0.9.11 - not very sure, but think it was 0.9.11 in my last tests) Instead it now behaves like routines, wich have to concat Zero-Terminated Strings. Nothing changed in my wine-configuration - only apt-getted to 0.9.12 first -> then to the latest 0.9.16 (ubuntu-deb) -> same problem. Do you handle the OLE-BSTR-stuff directly in wine, or do you give that jobs to other Libs in the system (the 0.9.12er winelib- update has also changed some dependencies (Gtk, libc* and others)? Olaf Schmidt

June 24, 2006

Re: indented relay traces

by Jaap Stolk

On 6/24/06, James Hawkins <truiken(a)gmail.com> wrote: > Ah, didn't know about that tool. You learn something new every day. didn't know about that tool too. sounds interesting. usually I do something like: // reduce log size: egrep -v 'RtlFreeHeap|RtlAllocateHeap|trace:syslevel:_Enter|trace:syslevel:_Leave|Ret kernel32.MultiByteToWideChar|Ret ntdll.RtlUpperChar' all.txt > all2.txt // add line numbering: (to back-trace interesting things) all2.txt > all.txt grep -n '' // filter out interesting things: (manually add more interesting finds later) egrep ':loaddll:|:ntoskrnl:|:warn:|Unable to|fixme' all.txt > all_interesting.txt Jaap.

June 24, 2006

Re: Autocad 2004 STATUS_INVALID_LDT_OFFSET

by Jaap Stolk

On 6/24/06, Jaap Stolk <jwstolk(a)gmail.com> wrote: > here is the log after remofing the tag altogether: > (I had to ctrl-c wine, autocad did not shutdown by itself, so I think > things are improving.) > > $ grep -i ':ntoskrnl:' all_log_no_tag.txt > 000f:trace:ntoskrnl:ExAllocatePoolWithTag (1 16 <tag>), semi-stub > 000f:trace:ntoskrnl:ExFreePool (0x401fcc18) It looks like something is randomly failing here. I just created a new +all log file, but this time there is no reference to ntoskrnl at all. I'm going to run a full memory check just to rule things out :-)

June 24, 2006

Re: indented relay traces

by James Hawkins

On 6/24/06, Eric Pouech <eric.pouech(a)wanadoo.fr> wrote: > > > this won't work for a multithreaded program > tools/examine_relay does what you want, plus some other goodies (calls > that didn't return...) Ah, didn't know about that tool. You learn something new every day. > IMO it's better to instrument the tools after the relay trace, rather > than tweaking the trace generation for any potential goodie > It's ok, I know it was a hack :) -- James Hawkins

June 24, 2006

Re: indented relay traces

by Eric Pouech

James Hawkins wrote: > Hey guys, > > I don't know if you've ever wished the relay traces were indented by > the level of the call, so you can get a better feel for who calls what > from where, but I thought it would be nifty. I've attached a patch > for dlls/ntdll/relay.c that gives you this. You see the call level > next to the thread id, then twice that number of spaces, then the > usual relay information. I'd paste an example, but it would just get > line wrapped, so patch your tree and give it a shot! > this won't work for a multithreaded program tools/examine_relay does what you want, plus some other goodies (calls that didn't return...) IMO it's better to instrument the tools after the relay trace, rather than tweaking the trace generation for any potential goodie A+

June 24, 2006

Re: setupapi: Fix SetupCloseInfFile when a NULL handle is given, with tests

by Joris Huizer

James Hawkins wrote: > Hi, > > This fixes bug 5511. > > Changelog: > * Fix SetupCloseInfFile when a NULL handle is given, with tests. > > dlls/setupapi/parser.c | 2 ++ > dlls/setupapi/tests/parser.c | 9 +++++++++ > 2 files changed, 11 insertions(+), 0 deletions(-) > > -- > James Hawkins > > > ------------------------------------------------------------------------ > > This attachment seems to be empty; regards, Joris

June 24, 2006

indented relay traces

by James Hawkins

Hey guys, I don't know if you've ever wished the relay traces were indented by the level of the call, so you can get a better feel for who calls what from where, but I thought it would be nifty. I've attached a patch for dlls/ntdll/relay.c that gives you this. You see the call level next to the thread id, then twice that number of spaces, then the usual relay information. I'd paste an example, but it would just get line wrapped, so patch your tree and give it a shot! -- James Hawkins

June 24, 2006

Re: How do I get the unix filename for a wine handle?

by Alexandre Julliard

Christoph Probst <chris-usenet(a)netzpunkt.org> writes: > My plan is to hook in NtWriteFile() (dlls/ntdll/file.c), because whenever a > windows program writes to a file this function is called. Within this > function the file is accessed using its unix_handle. > > The problem is that I need to know the unix filename for clamAV to access the > file, but I see no proper way or function to get it as long as I just know > the unix_handle or fd. You can't do that in general. In Unix a file can have multiple names, or even none at all, there's simply no way to get a filename from a handle. On Linux you can use /proc/self/fd but that's not very portable. -- Alexandre Julliard julliard(a)winehq.org

June 24, 2006

Re: Variable sized server replies

by Alexandre Julliard

"Thomas Kho" <tkho(a)ucla.edu> writes: > In protocol.def, the get_startup_info request declares two VARARGS as > replies. Unless I've mistaken, replies are limited to one VARARG that > the server call writes into preallocated data, right? There's one block of data, but it can contain multiple things, that's why you can have multiple VARARGs as long as there's a way to determine their length. > I can think of two ways to tackle this: > > 1. Do a server call before hand to request the size of the data to be > returned. This has the overhead of the additional call, but doesn't > increase the complexity of the server itself. > > 2. Instead of preallocating memory for the variable reply, have the > request mechanism allocate it (like it does on the server end for > variable sized data in the request). This also adds complexity as the > reply would be callee-allocated, caller-freed (or perhaps we could > take advantage of SERVER_END_REQ?). However, it seems kind of natural > to have the request process parallel the reply process. You cannot allocate memory while making a server call. Allocating memory on the client requires locking and possibly waiting on other threads, and you can't do that in the middle of a server call. > Is there another alterative? Any comments on which approach is better? The standard way is to allocate a reasonably sized buffer, perform the call, and if it returns a buffer overflow increase the allocated size and retry. Check the get_message call for an example. -- Alexandre Julliard julliard(a)winehq.org

June 24, 2006

Re: Autocad 2004 STATUS_INVALID_LDT_OFFSET

by Jaap Stolk

On 6/24/06, Vitaliy Margolen <wine-devel(a)kievinfo.com> wrote: > You can ignore tag altogether. It's only enabled with a special registry key (on > windows). And drivers have no way nor need to verify the tag. So I think you can > just remove that trace. I will need to think something to stop of from crashing. > Or try using wine_dbgstr_a((char*)&Tag) instead. ok. I have not made a real patch, but but if it pops up again i may try that. > PS: I wasn't sure if you were using my ntoskrnl implementation or not. yes. I'm using: http://www.winehq.com/pipermail/wine-devel/2006-June/048583.html > If you are then could you post somewhere +ntoskrnl trace? That might be helpful. > Also it could be number of other things that this driver tries to do that ntoskrnl > has not implementation for. It seems to work ok with autocad 2004. I think the problems are more in the safedisc itself. I have the +all log: (16 MB) $ grep -i ':ntoskrnl:' all_log.txt 163439:0009:trace:ntoskrnl:NTOSKRNL_connect Connecting to ntoskrnl 163444:0009:trace:ntoskrnl:NTOSKRNL_connect Trying to start ntoskrnl 178596:wine_dbg_vprintf: debugstr buffer overflow (contents: '000f:trace:ntoskrnl:ExAllocatePoolWithTag (0 20 ') 200646:0009:trace:ntoskrnl:NTOSKRNL_connect Connecting to ntoskrnl I just noticed, after my "fix" the there are no ntoskrnl traces in the +all log. so my fix was preventing the error by crashing ntoskrnl :-( here is the log after remofing the tag altogether: (I had to ctrl-c wine, autocad did not shutdown by itself, so I think things are improving.) $ grep -i ':ntoskrnl:' all_log_no_tag.txt 000f:trace:ntoskrnl:ExAllocatePoolWithTag (1 16 <tag>), semi-stub 000f:trace:ntoskrnl:ExFreePool (0x401fcc18) 000f:trace:ntoskrnl:ExAllocatePoolWithTag (1 94 <tag>), semi-stub 000f:trace:ntoskrnl:ExFreePool (0x4020d220) 000f:trace:ntoskrnl:ExAllocatePoolWithTag (0 168 <tag>), semi-stub 000f:trace:ntoskrnl:ExAllocatePoolWithTag (0 20 <tag>), semi-stub 000f:trace:ntoskrnl:ExAllocatePoolWithTag (0 120 <tag>), semi-stub 000f:trace:ntoskrnl:IoCreateDevice (0x4020d220, 0, L"\\Device\\CdaC17BA", 51712, 0, 0, 0x407523a0) 000f:trace:ntoskrnl:ExAllocatePoolWithTag (0 24 <tag>), semi-stub 000f:trace:ntoskrnl:ExAllocatePoolWithTag (0 184 <tag>), semi-stub 000f:trace:ntoskrnl:IoCreateDevice Created device with handel 0x28 000f:trace:ntoskrnl:IoCreateSymbolicLink (L"\\DosDevices\\CdaC17BA" -> L"\\Device\\CdaC17BA") 000f:trace:ntoskrnl:ExAllocatePoolWithTag (0 24 <tag>), semi-stub 000f:trace:ntoskrnl:ExAllocatePoolWithTag (1 36 <tag>), semi-stub 000f:trace:ntoskrnl:IoCreateDevice (0x4020d220, 0, L"\\Device\\CdaC15BA", 61184, 0, 0, 0x40752140) 000f:trace:ntoskrnl:ExAllocatePoolWithTag (0 24 <tag>), semi-stub 000f:trace:ntoskrnl:ExAllocatePoolWithTag (0 184 <tag>), semi-stub 000f:trace:ntoskrnl:IoCreateDevice Created device with handel 0x34 000f:trace:ntoskrnl:IoCreateSymbolicLink (L"\\DosDevices\\CdaC15BA" -> L"\\Device\\CdaC15BA") 000f:trace:ntoskrnl:ExAllocatePoolWithTag (0 24 <tag>), semi-stub 000f:trace:ntoskrnl:driver_control returned from call with status 00000000 does "semi-stub" mean that this call is not fully implemented ? Jaap.

June 24, 2006

Re: [AppDb] stuff thats broken

by Chris Morgan

> > 5.) If you add a hyperlink in a note it is broken with makesafe() > > turned on. > > I'm not sure this has anything to do with makeSafe(). Right now makeSafe() > is effectively disabled and I've carefully reviewed the changes to > addAppNode.php but I can't see how any of the changes broke notes. > > On my local machine after previewing a note I'm seeing any url I put in > prefixed by the current page url. On the appdb when previewing everything > appears to be working correctly. > > I think this may be a legit bug with the injection changes and I'll fix it > before committing. > Just tested again here with the injection changes and hypertext(tested with a url) in notes looks good to me. Chris

June 24, 2006

Re: [AppDb] stuff thats broken

by Chris Morgan

On Friday 23 June 2006 9:12 pm, Tony Lambregts wrote: > So I was testing Chris's "query_parameters" patch and I found the > following things broken with current CVS. This note is mostly to archive > the issues I found than anything else. Most of these are minor problems > but they should be fixed. If you want to have a go at fixing any of the > issues feel free to do so. > > 1.) When I go to enter a new distribution in the testing data (new > application, new version or new testing data) I get the following error > > We found the following errors: > > * Please enter a distribution. > > Please correct them. > > This is probably due to the makesafe() patch. > Fixed. > I have no idea how long these next 3 have not worked. > > 2.) I also cannot add a new category under the "Main" category This is > minor but we should fix it > This is a side effect of the call to html_select(). I don't see any good way to fix it other than by moving category editing into the category class. We should also fix the formatting of the dropdown so we do something smart like: Parent Category |-Child category |---Child category (space) Another parent category etc, so its a dropdown list but you can at least see what is under what. This could go onto the todo if someone is interested. > 3.) Editing a category is broken all fields are blank > Fixed. > 4.) deleting a category is broken. I get this error at least > > *Database Error!* > Query: DELETE FROM appCategory WHERE catId = LIMIT 1 > You have an error in your SQL syntax; check the manual that corresponds > to your MySQL server version for the right syntax to use near 'LIMIT 1' > at line 3 > Fixed. > 5.) If you add a hyperlink in a note it is broken with makesafe() > turned on. > I'm not sure this has anything to do with makeSafe(). Right now makeSafe() is effectively disabled and I've carefully reviewed the changes to addAppNode.php but I can't see how any of the changes broke notes. On my local machine after previewing a note I'm seeing any url I put in prefixed by the current page url. On the appdb when previewing everything appears to be working correctly. I think this may be a legit bug with the injection changes and I'll fix it before committing. > 6.) Voting is broken (I don't care really) the whole thing is kind > meaningless right now anyway. but that is another subject. > Fixed. Chris

June 24, 2006

Re: Very large files

by Vijay Kiran Kamuju

Hi, I think its dependent on the filesystem and the kernel version you are running. What is the Redhat linux version you are using snd the kernel version you are using. Thanks, Vijay On 6/23/06, John Willis <jswillis93105(a)verizon.net> wrote: > > Howdy, > > I have a large VS 6.0 C++ program and I'm running it under WINE under Redhat > Linux. > Everything works fine, but it doesn't handle file sizes greater than 4gb > like it can running > directly under Windows XP. > I've tweaked everything that I can find that seems to deal with file sizes. > Is there a problem with WINE itself handling very large file sizes? > > Thanks in advance, > John > > > >

June 24, 2006

Re: Autocad 2004 STATUS_INVALID_LDT_OFFSET

by Vitaliy Margolen

Friday, June 23, 2006, 9:47:01 AM, Jaap Stolk wrote: > ExAllocatePoolWithTag( ) function in the file ntoskrnl_api.c : > I changed: > WINE_TRACE("(%d %ld %.4s), semi-stub\n", PoolType, NumberOfBytes, (char*)&Tag); > to: > WINE_TRACE("(%d %ld %4s), semi-stub\n", PoolType, NumberOfBytes, (char*)&Tag); > Don't know how decimals are supposed to work with a string, but it > didn't work for me. You can ignore tag altogether. It's only enabled with a special registry key (on windows). And drivers have no way nor need to verify the tag. So I think you can just remove that trace. I will need to think something to stop of from crashing. Or try using wine_dbgstr_a((char*)&Tag) instead. PS: I wasn't sure if you were using my ntoskrnl implementation or not. If you are then could you post somewhere +ntoskrnl trace? That might be helpful. Also it could be number of other things that this driver tries to do that ntoskrnl has not implementation for. Vitaliy.

June 24, 2006

Re: [AppDB] - protect sql insert statements from injection attacks

by Tony Lambregts

Chris Morgan wrote: > Here is the new version. I've switched to the fully expanded method of > writing out the sql. This is the same format used by several db wrapper > libraries, in prepared sql statements and is the recommended style under > c#/.net for queries. > > I've tested creating new users and distributions and submitting an > application. > I have tested the patch thoroughly and it looks good to me. -- Tony Lambregts

June 24, 2006

Re: Wine developers?

by Vitaliy Margolen

Friday, June 23, 2006, 9:51:28 AM, Kuba Ober wrote: > On Wednesday 21 June 2006 21:24, Vitaliy Margolen wrote: >> Wednesday, June 21, 2006, 11:41:32 AM, Andreas Mohr wrote: >> > Hi, >> > >> > On Wed, Jun 21, 2006 at 07:27:28PM +0200, Kai Blin wrote: >> >> I wouldn't necessarily judge the whole list by just two negative >> >> reactions. It's interesting to see that with my wine experience there >> >> might be jobs out there working on something similar. (Not that I'm >> >> interested right now, I'm still at university...) >> >> >> >> I'm sure you're aware that in every open source project, there's people >> >> who oppose commercial software development for philosophical reasons. >> > >> > "proprietary" please, not "commercial". HUGE difference. >> >> I was talking more about nature of the job. I'm sure that lots of Wine code >> can "leak" into commercial product and wise versa. > What's worse is that it already DID happen, ane the party involved didn't > think much of it (remember Poject David?) But this has little to do with the > matter at hand. > I doubt that Thinstall would really want to do anything illegal -- this woul,d > ruin their reputation, and would likely expose them to a copyright > lawsuit(s). We can only level accusations after they do wrong. Why should we > a priori think of them as guilty? > I think that while Vitaliy has some firm moral (and otherwise) viewpoints, the > Thinstall bashing of his was a bit much. > C'mon, they just politely asked if anyone is interested in a job. What's wrong > with that? If I were looking for winapi hackers, I'd ask here myself. Jeez. I'm sorry that my intentions were misinterpreted because of the bad choice of words on my part. I didn't want to sound that negative. What I wanted to say, is that the nature of the Wine (windows internals) and Thinstall (internal knowledge of windows) are some what close. So it's kind of a thin line that I'm sure neither side wants to cross. As for the job offers - I wish you good luck finding the right person for the job. And for a Wine hacker to finally be able to pay his/hers bills. IMHO the project's developer's mailing list is not the right place to send such offers. PS: Again this is all IMHO and in no way should represent opinions of anyone else. Vitaliy.

June 24, 2006

[AppDb] stuff thats broken

by Tony Lambregts

So I was testing Chris's "query_parameters" patch and I found the following things broken with current CVS. This note is mostly to archive the issues I found than anything else. Most of these are minor problems but they should be fixed. If you want to have a go at fixing any of the issues feel free to do so. 1.) When I go to enter a new distribution in the testing data (new application, new version or new testing data) I get the following error We found the following errors: * Please enter a distribution. Please correct them. This is probably due to the makesafe() patch. I have no idea how long these next 3 have not worked. 2.) I also cannot add a new category under the "Main" category This is minor but we should fix it 3.) Editing a category is broken all fields are blank 4.) deleting a category is broken. I get this error at least *Database Error!* Query: DELETE FROM appCategory WHERE catId = LIMIT 1 You have an error in your SQL syntax; check the manual that corresponds to your MySQL server version for the right syntax to use near 'LIMIT 1' at line 3 5.) If you add a hyperlink in a note it is broken with makesafe() turned on. 6.) Voting is broken (I don't care really) the whole thing is kind meaningless right now anyway. but that is another subject. -- Tony Lambregts

June 24, 2006

Re: LoadResource with module=0x00400000 followed by stack overflow

by Jonathan Clark

On Tue, 20 Jun 2006 12:49:25 +0200, Jan-Espen Pettersen wrote: > Can a win32 program expect that it has write access to the memory > pointed to by the return value of LoadResource with module 0x00400000? > Should LoadResource in some cases copy the memory and return a > writeable copy of the resource? I noticed a similar problem under Windows Server 2003 when DEP is enabled, but haven't seen it on other platforms. If I recall, there is code in version.dll which calls LoadResource with RT_VERSION and then expects to be able to write to pages that are clearly marked not-writable in the PE image. The pages remain read-only until after a call to GetVersionInfo so something in the kernel is altering them. I didn't dig into it, but thought it was really weird. When I turn off DEP for that program I got different behavior. I don't know what this means for wine but maybe that is useful. Jonathan

June 23, 2006

Very large files

by John Willis

Howdy, I have a large VS 6.0 C++ program and I'm running it under WINE under Redhat Linux. Everything works fine, but it doesn't handle file sizes greater than 4gb like it can running directly under Windows XP. I've tweaked everything that I can find that seems to deal with file sizes. Is there a problem with WINE itself handling very large file sizes? Thanks in advance, John

June 23, 2006

How do I get the unix filename for a wine handle?

by Christoph Probst

Hi, I'm one of the Google Summer of Code students and I'm working on the ClamAV integration in wine. Today I was discussing an issue with my mentor and he suggested to post it here. Currently I'm working on a scan-after-write functionality: Whenever a file was changed the virusscanner checks the file. My plan is to hook in NtWriteFile() (dlls/ntdll/file.c), because whenever a windows program writes to a file this function is called. Within this function the file is accessed using its unix_handle. The problem is that I need to know the unix filename for clamAV to access the file, but I see no proper way or function to get it as long as I just know the unix_handle or fd. Possible solutions so far: 1. searching /proc/self/fd/%d for the name. This would be an easy but very ugly solution. 2. maintaining a table mapping unix_handles to filenames from within NtCreateFile(). Hence I don't have the feeling that this is a good solution. 3. storing the unix filename in the wineserver handle object when it's created. This is probably the right place to keep this information but it would require some changes to the wineserver. If I understand it right it is already planned to implement something like this ("lookup_name" in server/object.c) But maybe I just missed an even better approach. What would you recommend to do? Any comment is appreciated. Cheers, Chris -- Kontakt-Details: http://www.christoph-probst.com/kontakt/ PGP-FP: B171 7EA4 988C DD90 1601 D21C 5279 2FAF 9978 AF86

June 23, 2006

Variable sized server replies

by Thomas Kho

In protocol.def, the get_startup_info request declares two VARARGS as replies. Unless I've mistaken, replies are limited to one VARARG that the server call writes into preallocated data, right? That leads me to my next question: I'm currently working on moving menus into the server (see <http://silenceisdefeat.org/~tkho/seas/wine/patches/menu-0622-3/> for a series of patches that moves most menu data into the server. Most logic is still on the client, but it's still surprisingly snappy.) If you look at <http://silenceisdefeat.org/~tkho/seas/wine/patches/menu-0622-3/10-server-mii>, my problem is that I don't know the size of the data that the server is going to return and my current hack is to set a hard upper limit. I can think of two ways to tackle this: 1. Do a server call before hand to request the size of the data to be returned. This has the overhead of the additional call, but doesn't increase the complexity of the server itself. 2. Instead of preallocating memory for the variable reply, have the request mechanism allocate it (like it does on the server end for variable sized data in the request). This also adds complexity as the reply would be callee-allocated, caller-freed (or perhaps we could take advantage of SERVER_END_REQ?). However, it seems kind of natural to have the request process parallel the reply process. Is there another alterative? Any comments on which approach is better? Thomas Kho

June 23, 2006

Re: Wine developers?

by Kuba Ober

On Wednesday 21 June 2006 21:24, Vitaliy Margolen wrote: > Wednesday, June 21, 2006, 11:41:32 AM, Andreas Mohr wrote: > > Hi, > > > > On Wed, Jun 21, 2006 at 07:27:28PM +0200, Kai Blin wrote: > >> I wouldn't necessarily judge the whole list by just two negative > >> reactions. It's interesting to see that with my wine experience there > >> might be jobs out there working on something similar. (Not that I'm > >> interested right now, I'm still at university...) > >> > >> I'm sure you're aware that in every open source project, there's people > >> who oppose commercial software development for philosophical reasons. > > > > "proprietary" please, not "commercial". HUGE difference. > > I was talking more about nature of the job. I'm sure that lots of Wine code > can "leak" into commercial product and wise versa. What's worse is that it already DID happen, ane the party involved didn't think much of it (remember Poject David?) But this has little to do with the matter at hand. I doubt that Thinstall would really want to do anything illegal -- this woul,d ruin their reputation, and would likely expose them to a copyright lawsuit(s). We can only level accusations after they do wrong. Why should we a priori think of them as guilty? I think that while Vitaliy has some firm moral (and otherwise) viewpoints, the Thinstall bashing of his was a bit much. C'mon, they just politely asked if anyone is interested in a job. What's wrong with that? If I were looking for winapi hackers, I'd ask here myself. Jeez. Cheers, Kuba

June 23, 2006

Re: Autocad 2004 STATUS_INVALID_LDT_OFFSET

by Jaap Stolk

ExAllocatePoolWithTag( ) function in the file ntoskrnl_api.c : I changed: WINE_TRACE("(%d %ld %.4s), semi-stub\n", PoolType, NumberOfBytes, (char*)&Tag); to: WINE_TRACE("(%d %ld %4s), semi-stub\n", PoolType, NumberOfBytes, (char*)&Tag); Don't know how decimals are supposed to work with a string, but it didn't work for me.

June 23, 2006

Re: Autocad 2004 STATUS_INVALID_LDT_OFFSET

by Jaap Stolk

I think i found the cause further up in the log. 178591:000f:Starting process L"c:\\windows\\system32\\ntoskrnl.exe" ... 178596:wine_dbg_vprintf: debugstr buffer overflow (contents: '000f:trace:ntoskrnl:ExAllocatePoolWithTag (0 20 ') 178597:000f:trace:seh:raise_exception code=80000101 flags=1 addr=0xffffe002 ... I will look into it some more.

June 23, 2006

Re: Autocad 2004 STATUS_INVALID_LDT_OFFSET

by Jaap Stolk

On 6/23/06, Mike Hearn <mike(a)plan99.net> wrote: > On Fri, 23 Jun 2006 14:13:18 +0200, Jaap Stolk wrote: > > 200145:Backtrace: > > Hmm seems like you are seeing a mixed debugger/app trace there ... gotta > be careful of that! Ah, yes that's it. it seems one of the threads is bailing out somewhere. further up in he log: the *signal* starts: 198468:0011: get_key_value() = OBJECT_NAME_NOT_FOUND { type=-1, total=0, data={} } 198469:0011: close_handle( handle=0x28 ) 198470:0011: close_handle() = 0 { fd=-1 } 198471:0011: set_console_input_info( handle=(nil), mask=2, active_sb=(nil), history_mode=0, history_size=0, edition_mode=0, title=L"Wine Debugger" ) 198472:0011: set_console_input_info() = INVALID_PARAMETER 198473:0011: debug_process( pid=000e, attach=1 ) 198474:000f: *signal* signal=10 198475:000f: *signal* signal=19 198476:0011: debug_process() = 0 198477:000f: set_thread_context( handle=0xfffffffe, flags=00010007, suspend=1, context={flags=0001001f,eax=00000003,ebx=00000007,ecx=4074efc0,edx=00000008,esi=400321d4,edi=4074efc0,ebp=4074efd8,eip=ffffe000,esp=4074ef94,eflags=00000246,cs=0023,ds=002b,es=002b,fs=003b,gs=0033,dr0=00000000,dr1=00000000,dr2=00000000,dr3=00000000,dr6=00000000,dr7=00000000,float={ffff037f,ffff0120,ffffffff,00000000,00000000,00000000,00000000,00000000,00000000,00000000,00000000,00000000,00000000,00000000,00000000,00000000,00000000,00000000,00000000,00000000,00000000,00000000,00000000,00000000,00000000,22a00000,401d8e7c,00000120}} ) 198478:000f: set_thread_context() = 0 { self=1 } still wondering why it didn't do that yesterday.

June 23, 2006

Re: Autocad 2004 STATUS_INVALID_LDT_OFFSET

by Mike Hearn

On Fri, 23 Jun 2006 14:13:18 +0200, Jaap Stolk wrote: > 200145:Backtrace: Hmm seems like you are seeing a mixed debugger/app trace there ... gotta be careful of that!

June 23, 2006

Re: Move all common Dll[un]RegisterServer stuff to advapi

by Mike Hearn

On Thu, 22 Jun 2006 15:58:58 -0700, James Hawkins wrote: > The imports for current dlls concerning advapi32 are fine; no need to > change to delay import. I understand where you're going with moving > factoring out registration, it's just that advapi32 is not the place. > I'm not sure where else you could put this. I would ask Alexandre > whether he wants it done at all, and if so, where the new code should > be located. I believe this was discussed when the registration code was first added, and Alexandre preferred this approach to having some utility DLL. I guess it's a matter of taste. I would have preferred a wineutils.dll or somesuch but he is the boss :) thanks -mike

June 23, 2006

Re: Wine developers?

by Mike Hearn

On Tue, 20 Jun 2006 19:32:11 -0600, Vitaliy Margolen wrote: > Basically you are _stealing_ developers from the project. Because with your > closed source project such developer will be prohibited from participating in > the Wine project. A very specific legal interpretation that would require the company behind Thinstall to want to hurt the Wine project .... be careful, none of us are lawyers here.

June 23, 2006

Re: Wine developers?

by Mike Hearn

Hiya Jonathon, I wouldn't worry too much about the negative reactions there, which is a shame. As Molle has pointed out he is not really a Wine developer. I am and I'd say that it's totally fine to post such a job advert here, I'm sure there are people here who would like to find a good job with their skills. I hope you fill the position soon! Good luck! thanks -mike On Tue, 20 Jun 2006 13:59:59 -0700, Jonathan Clark wrote: > My name is Jonathan Clark, and I work with a team on a project that has > some similarities with Wine. The project is called Thinstall > (http://thinstall.com) and on first glance similarities may not be > apparent.

June 23, 2006

Re: LoadResource with module=0x00400000 followed by stack overflow

by Mike Hearn

On Tue, 20 Jun 2006 12:49:25 +0200, Jan-Espen Pettersen wrote: > Can a win32 program expect that it has write access to the memory > pointed to by the return value of LoadResource with module 0x00400000? > Should LoadResource in some cases copy the memory and return a writeable > copy of the resource? The SEH handling code should catch this case and un-protect the resource section. At least it does for InstallShield ... I'm not sure why it's not working here. thanks -mike

June 23, 2006

Autocad 2004 STATUS_INVALID_LDT_OFFSET

by Jaap Stolk

I'm trying to run Autocad 2004. (installation from cd "works") current status: (/blog) http://wiki.jswindle.com/index.php/AutoCad#Forum_Comments It starts a long way, and wine seems to handle the usual CreateFileA("\\\\.\\NTICE") and IsDebuggerPresent() properly, and it generates and runs some dll's in the temp directory. But the program exists without any waring or error. I'm still searching for the reason, but i have several clues. But today the log files are shorter, the above calls are missing, and instead I get an c0000119 error (STATUS_INVALID_LDT_OFFSET) 200135:0011: read_process_memory() = 0 { data={00,00,00,00} } 200136:0011: get_handle_fd( handle=0xc, access=00000002 ) 200137:0011: get_handle_fd() = 0 { fd=9, removable=0, flags=0 } 200138:0x4074fc00: 00000000 00000000 00000000 00000000 200139:0011: get_selector_entry( handle=0x2c, entry=7 ) 200140:000f: *signal* signal=19 200141:0011: get_selector_entry() = 0 { base=00000000, limit=00000000, flags=00 } 200142:fixme:ntdll:RtlNtStatusToDosErrorNoTeb no mapping for c0000119 200143:0011: get_handle_fd( handle=0xc, access=00000002 ) 200144:0011: get_handle_fd() = 0 { fd=9, removable=0, flags=0 } 200145:Backtrace: 200146:0011: get_handle_fd( handle=0xc, access=00000002 ) 200147:0011: get_handle_fd() = 0 { fd=9, removable=0, flags=0 } 200148:=>1 0xffffe002 (0xffffe002) 200149:0009:Ret ntdll.NtDelayExecution() retval=00000000 ret=4037416a 200150:0009:Ret kernel32.Sleep() retval=00000000 ret=408fc345 200151:0009:Call ntdll.NtQuerySystemInformation(0000000b,00000000,00000000,4072ef3c) ret=408fc296 I don't have any "+all" log files from yesterday, but I'm sure there was no "fixme" then. Is Wine actually simulating the LDT stuff ? I think I should try with a fresh install.

June 23, 2006

Re: ddraw assert prevents Ankh from starting

by Christoph Frick

On Fri, Jun 23, 2006 at 11:49:24AM +0200, Stefan Dösinger wrote: Hi Stefan, > I tested the demo, the loading screen appears without the assert. I will send > a more proper implementation of IDirectDrawSurface::SetSurfaceDesc when I > have time(university tests :-( ). no hurry! i try out all the stuff that lies on my disk with cvs every now and then and this came up. > The loading screen appears, but it seems to get stuck quite soon > during loading. It seems to wait for something, the cpu is idle and > the screen not redrawn. it takes _ages_ in some c++ code in one of the dlls it ships with. you just have to be patient. it takes about 3-5 minutes - then you are in the main menu (where the mouse is magically drawn to the center of the screen like a moth and yesterdays dinput mouse fix, does not help here). one good thing about it, is the use of ogre, so i had a peak in the dinput code of current ogre version, if i see something strange. but with my limited knowledge of if the winapi there was at least nothing obvious and there is always the possibility, that the old version did something radical else or that they have patched around in the ogre code. > This game seems to use the open source ogle library for rendering, and > it also ships a Direct3D9 and OpenGL renderer. Maybe there's a way to > make it use the ogl renderer, but I haven't found a way to do so. forget about that - it does not run at all using ogl. i had the same idea and tried it. its also a pity, that they use ogre but do not provide a osx/linux version of the game :/ -- cu

June 23, 2006

Re: ddraw assert prevents Ankh from starting

by Stefan Dösinger

Am Dienstag 20 Juni 2006 09:17 schrieb Christoph Frick: > hiho > > on the end of this mail is a patch, that removes an assert(0) from the > surface code in ddraw. but removing it let Ankh[1][2] start (it started > before, but there where no loading screen visible). yet there are other > problems with this game - but it now actually works better than before. Oh, just noticed that ankh renders with direct3d9 by default. It can be switched to opengl by commenting out the d3d9 plugin in bin/release/plugins.cfg and enablding the opengl plugin instead. However, this doesn't fix the loading bug, and it still hits the assertion in ddraw.

June 23, 2006

Re: ddraw assert prevents Ankh from starting

by Stefan Dösinger

Am Dienstag 20 Juni 2006 09:17 schrieb Christoph Frick: > hiho > > on the end of this mail is a patch, that removes an assert(0) from the > surface code in ddraw. but removing it let Ankh[1][2] start (it started > before, but there where no loading screen visible). yet there are other > problems with this game - but it now actually works better than before. I tested the demo, the loading screen appears without the assert. I will send a more proper implementation of IDirectDrawSurface::SetSurfaceDesc when I have time(university tests :-( ). The loading screen appears, but it seems to get stuck quite soon during loading. It seems to wait for something, the cpu is idle and the screen not redrawn. This game seems to use the open source ogle library for rendering, and it also ships a Direct3D9 and OpenGL renderer. Maybe there's a way to make it use the ogl renderer, but I haven't found a way to do so.

June 23, 2006

Re: riched20: EM_GETLINE implementation and test

by Andrew Talbot

Marcus Meissner wrote: > In Linux, L"" makes 4 byte character strings (if not using explicit > -fshort-wchar). Also, if the user then calls wc* functions from glibc, > they will use 4 byte characters. > > It is all to avoid confusion. > > Ciao, Marcus Interestingly, it looks like, in one standard version of C, at least - maybe not C89 - there is now a pair of utf16 string-literal constructs: u'x' an u"string". But that's another story. Thanks, Marcus (and Mike)! -- Andy.

June 23, 2006

Re: [AppDB] - protect sql insert statements from injection attacks

by Chris Morgan

Here is the new version. I've switched to the fully expanded method of writing out the sql. This is the same format used by several db wrapper libraries, in prepared sql statements and is the recommended style under c#/.net for queries. I've tested creating new users and distributions and submitting an application. Chris On Friday 23 June 2006 1:05 am, Chris Morgan wrote: > On Thursday 22 June 2006 10:21 pm, Tony Lambregts wrote: > > Chris Morgan wrote: > > > Change compile_insert_string() to compile_insert_array() and add a new > > > first parameter of $sTable that represents the table you are building > > > the insert statement for. The output of compile_insert_array() is > > > passed to query_insert() to perform the insertion. Change all existing > > > calls of compile_insert_string() to compile_insert_array() and specify > > > the table as the first parameter. Make all instances of sql inserts > > > that didn't use compile_insert_string() use compile_insert_array(). > > > > > > The use of mysql_real_escape_string() inside of compile_insert_array() > > > protects the insert statement from sql injection attacks. > > > > > > Chris > > > > Just a couple of nits > > > > This Could have been broken before but when I try to edit a bundle I > > get: *Database Error!* > > Query: SELECT bundleId, appBundle.appId, appName FROM appBundle, > > appFamily WHERE bundleId = AND appFamily.appId = appBundle.appId > > You have an error in your SQL syntax; check the manual that corresponds > > to your MySQL server version for the right syntax to use near 'AND > > appFamily.appId = appBundle.appId' at line 1 > > This is fixed now. Was broken all the way back to the import into cvs :-) > > > This is probably broken from before (makesafe?) but I cannot create a > > new distribution. > > Yes, makesafe changes broke this. Fixed now. > > > Here is a big one though. I cannot create a new account. I just get a > > blank screen. This is .new see below > > > > You could have broken the patch in two IE: > > > > Change log: The use of mysql_real_escape_string() on inserts to protect > > from sql injection attacks. > > > > Files changed: > > include/application.php / > > include/bugs.php / > > include/category.php / > > include/comment.php / > > include/db.php / > > include/distributions.php > > include/monitor.php / > > include/note.php / > > include/screenshot.php > > include/testResults.php / > > include/url.php / > > include/user.php / > > include/util.php / > > include/vendor.php / > > include/version.php / > > > > Change log: Make all remaining instances of sql inserts use > > compile_insert_array(). > > > > Files changed: > > maintainersubmit.php / > > admin/adminAppDataQueue.php / > > admin/editBundle.php / > > include/appdb.php / > > include/vote.php / > > > > I'm not say you should have just that it could have been done that way. > > > > You should include the "change log:" and "files changed:" to the patch > > the files. "Changed files: really helps in that it gives you an > > overview of the scope of the patch. > > > > > ----------------------------------------------------------------------- > > >- > > > > > > ? compile_insert_array.patch > > > ? hits_table_alter > > > ? injection_protect.patch > > > ? injection_protect2.patch > > > ? limittestresults.patch4 > > > ? vote_table_alter > > > ? data/screenshots > > > > Last really trivial nit (these are not really part of the patch) > > > > > Index: include/user.php > > > =================================================================== > > > RCS file: /opt/cvs-commit/appdb/include/user.php,v > > > retrieving revision 1.67 > > > diff -u -r1.67 user.php > > > --- include/user.php 21 Jun 2006 01:04:13 -0000 1.67 > > > +++ include/user.php 22 Jun 2006 20:07:17 -0000 > > > @@ -83,14 +83,15 @@ > > > return false; > > > } else > > > { > > > - $aInsert = compile_insert_string(array( 'realname' => > > > $sRealname, > > > - 'email' => > > > $sEmail, - > > > 'CVSrelease' => $sWineRelease )); > > > + $aInsert = compile_insert_array("user_list", > > > + array( 'realname' => > > > $sRealname, > > > + 'email' => > > > $sEmail, > > > + 'CVSrelease' => > > > $sWineRelease, > > > + 'password' => > > > password($sPassword), > > > + 'stamp' => > > > "NOW()", > > > + 'created' => > > > "NOW()")); > > > > I remember fighting with this before. password() and NOW() cannot be > > put into the array like this I do not have an easy answer for this at > > all at this point > > > > > - $sFields = "({$aInsert['FIELDS']}, `password`, `stamp`, > > > `created`)"; > > > - $sValues = "({$aInsert['VALUES']}, > > > password('".$sPassword."'), NOW(), NOW() )"; > > > - > > > - query_appdb("INSERT INTO user_list $sFields VALUES > > > $sValues", "Error while creating a new user."); > > > + query_insert($aInsert, "Error while creating a new > > > user."); > > > > > > $retval = $this->login($sEmail, $sPassword); > > > $this->setPref("comments:mode", "threaded"); /* set the > > > users default comments:mode to threaded */ > > > @@ -183,7 +184,11 @@ > > > return false; > > > > > > $hResult = query_appdb("DELETE FROM user_prefs WHERE userid = > > > ".$this->iUserId." AND name = '$sKey'"); > > > - $hResult = query_appdb("INSERT INTO user_prefs > > > VALUES(".$this->iUserId.", '$sKey', '$sValue')"); > > > + $aInsert = compile_insert_array("user_prefs", > > > + array( 'userid' => > > > $this->iUserId, > > > + 'name' => $sKey, > > > + 'value' => $sValue)); > > > + $hResult = query_insert($aInsert); > > > return $hResult; > > > } > > > > Other than that its clean. We need to find a way of fixing new > > accounts though. > > > > -- > > > > Tony Lambregts > > I've got a fix for this. The basic jist is to simply get rid of > compile_insert_array() and use a fancier version of query_parameter() that > uses pear db replacement operators. There isn't any other way to know > whether any given variable should be surrounded with '' or not > escaped(something else that is useful in certain cases apparently). Should > have this change in place soon. > > Chris

June 23, 2006

CFileDialog::OnFolderChange() problem with Customized file open dialog

by Dan Kegel

Ravinder wrote: > I am currently working on making a windows program function in > wine. it works for the most part, so i'm trying to iron out a > few critical bugs. first off... i'm looking at microsoft's > CFileDialog class - it is what the standard windows 'open > file' and 'save file' dialog windows use. the problem is this: > CFileDialog has a built in function called OnFolderChange, that > is used to determine whether the user has changed folders in the > dialog, whether by double click or what not.. the program i'm > working on overrides the CFileDialog class, and includes more > code for the OnFolderChange function (and for other functions) > for its own purposes. the new OnFolderChange function works > perfectly fine in windows, but is completely ignored when using > wine - it never even gets called, while the rest of the dialog > window retains all of its intended functionality. I see you've filed http://bugs.winehq.org/show_bug.cgi?id=5475 about on Jonathan Ernst's advice. Good advice! But be sure to add the full source code and project file for your sample code, as some people do have Visual C++, and will want to build it and step through. ATL problems are, I suspect, hard to diagnose because it requires understanding a bit about how ATL works under the hood, which few people do. If you're in a hurry for a solution, you could engage Codeweavers to solve problems like this. - Dan -- Wine for Windows ISVs: http://kegel.com/wine/isv

June 23, 2006

Re: riched20: EM_GETLINE implementation and test

by Mike McCormack

James Hawkins wrote: > To be as portable as possible, we code to the lowest common > denominator. Not all compilers support L"". We adhere to C89, but I > can't remember what it says about L"", if anything. The point is that > WCHAR blah[] ={...} will work for every compiler. wchar_t is an int on some versions of gcc. It can be made a short using -fshort-wchar, but only later versions of gcc (3.0+) support that flag. Mike

June 23, 2006