September 2021 - Wine-Bugs - list.winehq.org

[Bug 50322] New: REDEngineErrorReporter.exe Page fault while trying to print unimplemented function message to console.
by WineHQ Bugzilla April 13, 2022

April 13, 2022

https://bugs.winehq.org/show_bug.cgi?id=50322 Bug ID: 50322 Summary: REDEngineErrorReporter.exe Page fault while trying to print unimplemented function message to console. Product: Wine Version: 6.0-rc2 Hardware: x86-64 OS: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: -unknown Assignee: wine-bugs(a)winehq.org Reporter: hibbsncc1701(a)gmail.com Distribution: --- Created attachment 68905 --> https://bugs.winehq.org/attachment.cgi?id=68905 Backtrace of bug. Crash occurs when attempting to debug print an unimplemented function call when launching Cyberpunk2077 (GOG version) using dxvk with wine's dxgi.dll. Triggered by REDEngineErrorReporter.exe and not the game itself. -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.

1 3

[Bug 45023] New: TestBot Object Relational Mapper issues
by wine-bugs＠winehq.org April 13, 2022

April 13, 2022

https://bugs.winehq.org/show_bug.cgi?id=45023 Bug ID: 45023 Summary: TestBot Object Relational Mapper issues Product: Wine-Testbot Version: unspecified Hardware: x86 OS: Linux Status: NEW Severity: normal Priority: P2 Component: unknown Assignee: wine-bugs(a)winehq.org Reporter: fgouget(a)codeweavers.com Distribution: --- The TestBot implements its own perl Object Relational Mapper (ORM) to access the database. It is implemented by the modules in lib/ObjectModel. The are two main concepts: * Items An Item is an object representing a table row. The classes such as Job, Step and Task all inherit from Item. * Collections A collection represents a table and is thus a collection of Items. To allow for fast access the items are stored in a perl hashtable indexed by their primary key. A Collection may not contain all of the table's rows. For instance a Step Collection only contains the Steps of the parent Job object used to create it. So $Job->Steps is a collection that contains only the $Job steps. The items in a Collection can be further filtered out by specifying that one of their property must be in a set of values. For instance $Jobs->AddFilter("Status", ["queued", "running"]) will cause the $Jobs collection to only return the queued and running jobs. Unfortunately this implementation has a number of limitations and design issues. 1. Care must be taken when creating multiple objects. my $Job1 = $Jobs->Add(); ... Set mandatory fields ... my $Job2 = $Jobs->Add(); ... Set mandatory fields ... # Here $Job1 != $Job2 $Jobs->Save(); Only $Job2 got saved to the database! The reason is that a Job's primary key is an auto-increment field. So the Job Id is set when it is saved to the database. Thus when created in memory that field is unset and an empty string is used when adding it to the $Jobs items hashtable. Thus $Job2 overwrites $Jobs->{Items}->{""} and is the only one that gets saved. Note that there is no such issue for Steps for instance because their key is not an auto-increment field and is set as soon as the object is created. Fortunately the workaround is simple: systematically save an object before creating the next one but the difference in behavior between $Jobs->Add() and $Steps->Add() is error prone. 2. Auto-increment values on new Items are not ignored If an auto-increment field is modified on an object, that value is used when saving the object to the database. This means fixing the previous issue is not as simple as setting $Job->Id(--$Unique): handling of auto-increment fields also needs to be changed when saving new objects. 3. Items retrieved from Detailref collections are missing fields 'Detailref' properties are used to represent 'one to many' relationships. They are used for instance to access the Steps that belong to a Job: my $Step = $Job->Steps->GetItem(1); Here Steps is a collection that gets created based on the information of the job's DetailRef property. However objects gotten in this way don't provide access to all of the corresponding database fields. In the case of Steps, although the Steps table has a JobId column, the $Step->JobId property does not exist. This presents difficulties in a number of cases such as when trying to figure out the name of the directory containing the Step's files since it is called "jobs/$JobId/$StepNo" (see Step::GetDir()). It also means there is no $Step->Job property to get back to the parent Job object (this would also cause a perl reference loop which would be troublesome). The reason for this issue is that when a Collection or Item has a parent object (called a 'master' object in the TestBot ORM) the parent's foreign key fields (Step.JobId here) are stored separately in a 'master cols' structure and are thus not accessible through the normal field methods. 4. "Child" objects can only be accessed from their parent Any collection referenced through a Detailref property can only be accessed through the corresponding 'parent' object. This is the case for the Steps collections. The only way to get a collection of steps is through $Job->Steps and this only returns the steps that belong to $Job. So it's not possible to iterate over all the rows in the Steps table. There is a (somewhat incorrect) workaround for this for the Records table. This involves having the CreateRecords() method picking one of two PropertyDescriptors list depending on the presence of a parent object. However the corresponding Record objects have an extra field compared to the ones obtained through $RecordGroup->Records: this time the foreign key identifying the parent object is accessible since it's not not tucked away in the 'master cols' structure. This is not an issue unless you start mixing both types of objects, for instance through scope inheritance. 5. The 'master cols' terminology is ambiguous. It's never clear if a 'master cols' method treats $self as a child of its parent object, or as the parent to other child objects. * $Step->GetMasterCols() treats $Step as a child object and returns the key columns of its $Job master object, that is (JobId). * Similarly $Step->MasterKeyChanged() means the parent object's key changed, so here that the value of $JobId changed (which happens on new objects due to the auto-increment field btw). * Despite the unchanged 'Master' moniker, $Step->GetMasterKey() considers $Step to be the master object of the Tasks it contains and thus includes the step's key in the returned columns. This means $Step->GetMasterKey() returns (JobId, StepNo). * Despite not having the 'Master' moniker, $Step->GetFullKey() works the same as $Step->GetMasterKey() but returns the column values as a single string '$JobId#@#$StepNo'. 6. Filters on Detailref collections It's possible to put a filter on a collection to only get a subset of the corresponding rows. For instance $Job->Steps->AddFilter('Status', ['queued']) ensures that one will only get the queued steps of $Job. But Detailref collections are stored as a field of the Item they belong to. This means $Job->Steps always returns the same collection object. So once a piece of code has set a filter on it, to only return 'queued' steps for instance, all other parts of the code will only get the 'queued' steps, no matter what they need. Also there is no method for cloning a collection (so one cannot do $Job->Steps->Clone()->AddFilter()), or for removing a filter. For instance: $Job->Steps->AddFilter('Status', ['queued']); ... # Here UpdateStatus() will only take into account the queued steps! $Job->UpdateStatus(); Note also that it's not entirely trivial to work around. It's tempting to do something like: $Job->Steps->AddFilter('Status', ['queued']); ... # Make sure the database contains an up-to-date view of all the Steps and # Tasks in $Job. $Job->Save(); # Then create a new job object from the database my $TmpJob = CreateJobs()->GetItem($Job->Id); # Now we can update the status without fearing bad filters. $TmpJob->UpdateStatus(); # But here $Job->Status could still be wrong so any code that still uses # $Job will be lead astray. So update $Job. $Job->Status($TmpJob->Status) # But that still leaves out every Step and Task object referenced by $Job # Also saving $Job will save $Job->Status, again. What could go wrong? 7. Filters vs. loading from the database Once a collection has been loaded from the database, adding a filter to it has no effect. That's because filtering happens by tweaking the SQL query and changing the filter does not mark the collection for reloading. Again this issue is made more severe because of the way Detailref collections are handled. If a piece of code has caused the $Job->Steps collection to be loaded, then adding a filter elsewhere will have no effect. For instance: $Job->UpdateStatus(); # calls $Job->GetItems(), loading it from DB ... $Job->Steps->AddFilter('Status', ['queued']); foreach my $Step ($Job->Steps->GetItem()) # Here we will get all steps, not just the queued ones! 8. No "or" operator in filters Originally the filters only allowed one to request that a field be in a set of admissible values. This was extended to allow requesting for a field to be less than, greater than or like some value. However it is still not possible to use an "or" operator in filters. So for instance one cannot retrieve all Steps that completed recently or correspond to a WineTest run (WHERE Ended > 5 minutes ago OR User == 'batch'). 9. No Itemref support for multiple key fields. A Task has a VMName field which is a foreign key referencing a VM. Through the use of an Itemref property one can directly access the VM with $Task->VM. But we cannot do the same thing for steps. The primary key of a step has two fields, (JobId, No). A step also has a PreviousNo field which identifies the step it depends on. So we would want to be able to access the previous step through $Step->Previous by using an Itemref property: CreateItemrefPropertyDescriptor("Previous", "Previous step", !1, !1, \&CreateSteps, ["JobId", "PreviousNo"]), However this fails with a "Multiple key components not supported" error. Note that for this to work we would also need a way to tell the Itemref that the Step.PreviousNo field should be mapped to Step.No. But there is no way to do so. 10. Collection::GetItem() blindly adds the item to the collection Collections have a GetItem() method that returns an item when given a string containing that object's key. So using the Step example above, if we still have the right $Job->Steps collection accessible, we can easily get the Previous step through: $Job->Steps->GetItem($Step->PreviousNo) However this also adds $Step->PreviousNo to the $Job->Steps collection, regardless of what the filter on that collection is. So when the scheduler analyzes the queued and running steps to figure out whether the one they depend on, $Step->PreviousNo, has completed, it also unwittingly adds the previous step to the $Job->Step collection. So the next iteration on the collection may return completed steps despite expectations. 11. Save order issues The TestBot ORM automatically takes care of saving parent objects before the child objects they contain. In practice, if you create a Job > Step > Task hierarchy you can count on the TestBot ORM to save the Job before saving the Steps that use the JobId as part of their key and so on. But this breaks down for $Step->PreviousNo. There you have to make sure to save the steps in the right order otherwise you get an SQL error. Similarly deletions must be performed in the right order. Fortunately this is mostly transparent since we only delete whole Jobs and Job::OnDelete() takes care of blanking the Step::PreviousNo fields before recursing. 12. No Order-by support There are a many cases where we retrieve a number of rows from the database and then do a simple alphabetical or numerical sort on them. That's the kind of thing that the database would do much faster than Perl. This is mostly an issue for the activity and statistics web pages because of the number of RecordGroups they handle. So it would be nice to be able to specify an Order-By SQL directive when retrieving the objects. However this would run into the same issues as the filters for Detailref collections with regards to already loaded collections, and GetItem() not knowing where to insert freshly loaded objects. 13. No support for joins There are a few cases where doing a join could be useful. For instance when reporting the activity what we really want is all the Records rows corresponding to a RecordGroup that is less than 12 hours old. For now we have to proceed indirectly: we query all the RecordGroup objects that are less than 12 hours old, take the lowest GroupId we find and then retrieve all the Records that have a Group Id greater than that. A different type of join could also be useful in many other places: currently we first retrieve the jobs and then we do one more SQL request per job to retrieve its steps and then one per step to retrieve the tasks. A join could let us load it all in one request. Fortunately we don't have many jobs so unlike in the activity case this does not have a significant performance impact. 14. Performance Most parts of the TestBot don't have much processing to do so that performance is not an issue (though there is not really any hard data). Things are different for the activity and statistics page. The statistics page is the worst and generating it takes over a dozen seconds. That's annoying because of the load it puts on the TestBot web server. It's also really long for a mere couple dozen thousand Records. Not all of it comes from the ORM but over 50% of it does. There are some relatively obvious paths for improvement. For instance accessing a field like $Job->Ended causes the ORM to loop over all of a job's property descriptors until it finds one called 'Ended', and then it returns the value from a hash table. It should be possible to make it so that most of the time the value is returned directly from the hash table. Detailref and Itemref properties have their own hash tables and so must be handled separately. But hat could likely also be changed. -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.

2 7

[Bug 51538] New: Control can no longer launch from an ntfs partition
by WineHQ Bugzilla April 13, 2022

April 13, 2022

https://bugs.winehq.org/show_bug.cgi?id=51538 Bug ID: 51538 Summary: Control can no longer launch from an ntfs partition Product: Wine-staging Version: 6.13 Hardware: x86-64 OS: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: -unknown Assignee: wine-bugs(a)winehq.org Reporter: ritalat(a)fastmail.com CC: leslie_alistair(a)hotmail.com, z.figura12(a)gmail.com Distribution: --- Created attachment 70377 --> https://bugs.winehq.org/attachment.cgi?id=70377 control console output I have a shared NTFS partition that I have used between Arch Linux and Windows 10 for a few years now. I have been able to play games such as Control in the past from it. Now multiple games fail to launch from that partition with similar "err:module:import_dll Library X not found" errors. Copying the entire game to a ext4 partition fixes the issue. -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.

1 11

[Bug 50526] New: ConEmu wineconsole Ctrl+m has different behavior
by WineHQ Bugzilla April 12, 2022

April 12, 2022

https://bugs.winehq.org/show_bug.cgi?id=50526 Bug ID: 50526 Summary: ConEmu wineconsole Ctrl+m has different behavior Product: Wine Version: 6.0 Hardware: x86-64 OS: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: inseng Assignee: wine-bugs(a)winehq.org Reporter: adrianekaleksander(a)gmail.com Distribution: --- Cmd on Windows: Ctrl+M moves the cursor to the next line (ENTER key) On wine: Ctrl+M shows ^M 1.Open wineconsole 2.Press ctrl+m -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.

1 5

[Bug 47287] New: Accessible Event Watcher Tool 'accevent.exe' from Windows 10 SDK (10.0.17763.x) crashes on unimplemented function oleacc.dll.AccessibleObjectFromEvent
by wine-bugs＠winehq.org April 12, 2022

April 12, 2022

https://bugs.winehq.org/show_bug.cgi?id=47287 Bug ID: 47287 Summary: Accessible Event Watcher Tool 'accevent.exe' from Windows 10 SDK (10.0.17763.x) crashes on unimplemented function oleacc.dll.AccessibleObjectFromEvent Product: Wine Version: 4.9 Hardware: x86-64 OS: Linux Status: NEW Severity: normal Priority: P2 Component: oleacc Assignee: wine-bugs(a)winehq.org Reporter: focht(a)gmx.net Distribution: --- Hello folks, as it says. Main menu 'Events', select 'Start Listening' and then click random main menu item again -> crash. --- snip --- $ pwd /home/focht/.wine/drive_c/Program Files (x86)/Windows Kits/10/bin/10.0.17763.0/x64 $ wine ./accevent.exe ... 0036:err:ntdll:NtQueryInformationToken Unhandled Token Information class 26! wine: Call from 0x7b45323c to unimplemented function oleacc.dll.AccessibleObjectFromEvent, aborting wine: Unimplemented function oleacc.dll.AccessibleObjectFromEvent called at address 0x7b45323c (thread 0036), starting debugger... Unhandled exception: unimplemented function oleacc.dll.AccessibleObjectFromEvent called in 64-bit code (0x000000007b45323c). Register dump: rip:000000007b45323c rsp:0000000000248940 rbp:0000000000248a10 eflags:00000202 ( - -- I - - - ) rax:0000000000248960 rbx:0000000080000100 rcx:0000000000248960 rdx:0000000000248980 rsi:0000000000248a50 rdi:0000000000248990 r8:0000000000000002 r9:0000000000248a40 r10:0000000000000000 r11:0000000000659550 r12:000000018000a058 r13:0000000000000000 r14:0000000180005cc0 r15:0000000000000001 ... Backtrace: =>0 0x000000007b45323c RaiseException+0x6c(code=<is not available>, flags=<is not available>, nbargs=<is not available>, args=<is not available>) [/home/focht/projects/wine/mainline-src/dlls/kernel32/except.c:86] in kernel32 (0x0000000000248a10) 1 0x00007f6170d48dad __wine_spec_unimplemented_stub+0x3c(module=<is not available>, function=<is not available>) [/home/focht/projects/wine/mainline-src/dlls/winecrt0/stub.c:32] in oleacc (0x0000000000248a60) 2 0x00007f6170d230f7 __wine_stub_AccessibleObjectFromEvent+0x16() in oleacc (0x0000000000248b00) 0x000000007b45323c RaiseException+0x6c [/home/focht/projects/wine/mainline-src/dlls/kernel32/except.c:86] in kernel32: leaq 0xfffffffffffffff0(%rbp),%rsp 86 } Modules: Module Address Debug info Name (108 modules) ELF 7b400000- 7b86c000 Dwarf kernel32<elf> \-PE 7b420000- 7b86c000 \ kernel32 ELF 7bc00000- 7bd55000 Deferred ntdll<elf> \-PE 7bc20000- 7bd55000 \ ntdll ELF 7c000000- 7c004000 Deferred <wine-loader> PE 140000000- 14002f000 Deferred accevent PE 180000000- 18000d000 Deferred event ... ELF 7f617f6fc000- 7f617faa1000 Dwarf libwine.so.1 ELF 7f617faa3000- 7f617fccc000 Deferred ld-linux-x86-64.so.2 ELF 7ffd4edfc000- 7ffd4edfd000 Deferred [vdso].so Threads: process tid prio (all id:s are in hex) ... 00000035 (D) C:\Program Files (x86)\Windows Kits\10\bin\10.0.17763.0\x64\accevent.exe 00000038 0 00000036 0 <== --- snip --- --- snip --- $ winedump -j import event.dll | sed -n '/OLEACC/,/^$/p' offset 0000663c OLEACC.dll Hint/Name Table: 00007580 TimeDateStamp: 00000000 (Thu Jan 1 01:00:00 1970) ForwarderChain: 00000000 First thunk RVA: 000050C8 Thunk Ordn Name 000050c8 25 ObjectFromLresult 000050d0 27 WindowFromAccessibleObject 000050d8 4 AccessibleObjectFromEvent 000050e0 18 GetRoleTextW 000050e8 20 GetStateTextW --- snip --- Wine source: https://source.winehq.org/git/wine.git/blob/HEAD:/dlls/oleacc/oleacc.spec --- snip --- 2 @ stub AccessibleObjectFromEvent --- snip --- Microsoft docs: https://docs.microsoft.com/en-us/windows/desktop/api/oleacc/nf-oleacc-acces… $ sha1sum 17763.1.180914-1434.rs5_release_WindowsSDK.iso e702b5e5f2597d01eaee1eb1be7a34b0da0b6211 17763.1.180914-1434.rs5_release_WindowsSDK.iso $ du -sh 17763.1.180914-1434.rs5_release_WindowsSDK.iso 815M 17763.1.180914-1434.rs5_release_WindowsSDK.iso $ wine --version wine-4.9-95-g6d05fae6f7 Regards -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.

2 11

[Bug 42030] New: winedbg: Internal crash at 0x9f58fd40
by wine-bugs＠winehq.org April 12, 2022

April 12, 2022

https://bugs.winehq.org/show_bug.cgi?id=42030 Bug ID: 42030 Summary: winedbg: Internal crash at 0x9f58fd40 Product: Wine Version: 1.8.1 Hardware: x86 OS: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: winedbg Assignee: wine-bugs(a)winehq.org Reporter: kenorb(a)gmail.com Distribution: --- The following command crashes: $ winedbg --command < <(echo help) winedbg: Internal crash at 0x9f58fd40 Probably syntax isn't right, but still shouldn't internally crash. -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.

2 6

[Bug 51821] New: EVE Online Sisi Client Crashes Due to differences in how Wine and Windows handle LoadLibraryExW Undefined Behavior
by WineHQ Bugzilla April 12, 2022

April 12, 2022

https://bugs.winehq.org/show_bug.cgi?id=51821 Bug ID: 51821 Summary: EVE Online Sisi Client Crashes Due to differences in how Wine and Windows handle LoadLibraryExW Undefined Behavior Product: Wine Version: 6.16 Hardware: x86-64 OS: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: -unknown Assignee: wine-bugs(a)winehq.org Reporter: algebro(a)tuta.io Distribution: --- I've been chasing this for the last few days and think I may finally have enough information to finally open a bug report. As of a few days ago, the Sisi testing client for EVE stopped working in recent versions of Wine, and the logs showed an unhandled Python error in a module called launchdarkly_client.pyd: ```ImportError: Failed to load DLL: Success``` I chased this down in a debugger and noticed they're passing a relative path to LoadLibraryExW for this library only (full/absolute paths for every other module), and 8 for dwFlags: ```hLibModule = LoadLibraryExW((LPCWSTR)plVar4[3],(HANDLE)0x0,8)``` Per the Microsoft documentation, 8 for dwFlags is undefined when the first argument is a relative path: https://docs.microsoft.com/en-us/windows/win32/api/libloaderapi/nf-libloade… Weirdly enough, the library loads fine on Windows (I've confirmed in windbg), but it fails with error code 0x7e (Module not found) when trying to load it in wine. Other users running Wine 4.4 say Sisi launches fine and doesn't run into the error, although I havent been able to get older versions to build to try and test it, but it sounds like there is a difference between how recent Wine versions implement LoadLibraryExW() and how Windows does, at least with regards to the undefined behavior. Let me know if I can provide any other information that would be helpful for tracking this down. Thanks! -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.

1 35

[Bug 51220] New: version:info - The GetFileVersionInfoEx() checks fail in non-English locales
by WineHQ Bugzilla April 12, 2022

April 12, 2022

https://bugs.winehq.org/show_bug.cgi?id=51220 Bug ID: 51220 Summary: version:info - The GetFileVersionInfoEx() checks fail in non-English locales Product: Wine Version: 6.8 Hardware: x86-64 OS: Linux Status: NEW Severity: normal Priority: P2 Component: version Assignee: wine-bugs(a)winehq.org Reporter: fgouget(a)codeweavers.com Distribution: --- The GetFileVersionInfoEx() checks fail in non-English locales because the returned information seems to be localized even when they shouldn't be, i.e. for flags=0 and flags=FILE_VER_GET_NEUTRAL: https://test.winehq.org/data/patterns.html#version:info fr_FR: info.c:709: Test failed: [0] got 1036, expected lang is 1033 ja_JP: info.c:709: Test failed: [0] got 1041, expected lang is 1033 I suspect this is because this code block in GetFileVersionInfoSizeExW() systematically fails to find the English resource: LANGID english = MAKELANGID( LANG_ENGLISH, SUBLANG_DEFAULT ); hRsrc = FindResourceExW( hModule, MAKEINTRESOURCEW(VS_VERSION_INFO), (LPWSTR)VS_FILE_INFO, english ); I tried some alternative combinations for MAKELANGID() (though maybe not the right one) to no avail. So it's unclear to me if the issue is a bug in FindResourceExW() or an issue in the set up of the file version information in kernel32.dll. Note: * The fg-deb64-* tests are run in the fr_FR.UTF-8 locale. -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.

1 6

[Bug 32183] New: Cannot open console device read only, then read from it to get input
by wine-bugs＠winehq.org April 12, 2022

April 12, 2022

http://bugs.winehq.org/show_bug.cgi?id=32183 Bug #: 32183 Summary: Cannot open console device read only, then read from it to get input Product: Wine Version: 1.5.17 Platform: x86 OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: kernel32 AssignedTo: wine-bugs(a)winehq.org ReportedBy: us(a)edmeades.me.uk Classification: Unclassified Created attachment 42462 --> http://bugs.winehq.org/attachment.cgi?id=42462 CON test program (source and exe) (Email sent to wine-devel for assistance, but logging bug to track issue in case I do not get anywhere) CreateFile("\\.\CON", GENERIC_READ...) followed by ReadFile(...) works on windows and fails on wine. I need this particular sequence of events to work in order for a patch I have for cmd.exe to support CON input 'nicely' (read 'without hacks'). For simplicity sake I've cut this down to a tiny test program, which works on windows and fails on wine which does the following: Opens the device ("\\.\CON") with CreateFile with GENERIC_READ rights (which internally opens a CONIN$ device with the same access rights) Reads from the device with ReadFile - Because its a console device, this drops through to ReadConsoleW, which creates a CONOUT$ and then waits on a keystroke - Once the key is pressed (WCEL_Get) the key is 'inserted' into the input buffer by calling WriteConsoleInputW The issue is that WriteConsoleInputW requires GENERIC_WRITE access, but the CON device (\\.\CON) was opened as GENERIC_READ (and in fact fails if I try to open it with GENERIC_WRITE). CreateFile("CONIN$"...) will let me open in GENERIC_READ/GENERIC_WRITE mode and the program works on both windows and wine, but if you open CONIN$ GENERIC_READ only then it fails on wine and works on windows, with the same issue. Now on windows, this works... the question is how to make it work on wine... My gut feeling, with nothing backing this at all, is that WCEL_Get should not use WriteConsoleInputW to inject the values into the input buffer, instead making the server call directly, but passing through to the server call something to indicate that it is ok to add the data to the buffer, but I'm fast getting out of my depth! How to reproduce: Compile sample source (exe provided in zip)... Run as: "test 1" - this is the one I need to work... \\.\CON GENERIC_READ case "test 2" - this is a similar problem but opens CONIN$ GENERIC_READ "test 3" - this works on both windows and wine, opening CONIN$ GENERIC_READ|GENERIC_WRITE "test 4" - this fails on both windows and wine as you cannot open CON device with WRITE access Except for case 4, When its 'failing' - when you press a key in wine, it exits. When its 'working' it reads from keyboard, echos to screen until ctrl+Z (crtl+D on wine) is pressed and ends -- Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email Do not reply to this email, post in Bugzilla using the above URL to reply. ------- You are receiving this mail because: ------- You are watching all bug changes.

2 14

[Bug 51396] New: cl.exe sometimes fails with internal compiler error
by WineHQ Bugzilla April 12, 2022

April 12, 2022

https://bugs.winehq.org/show_bug.cgi?id=51396 Bug ID: 51396 Summary: cl.exe sometimes fails with internal compiler error Product: Wine Version: 6.12 Hardware: x86-64 OS: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: -unknown Assignee: wine-bugs(a)winehq.org Reporter: rpisl(a)seznam.cz Distribution: --- Cl.exe from MSVC 2019 Build Tools sometimes fails with: "fatal error C1001: Internal compiler error." A workaround is setting vcruntime140_1 to native. This is possibly caused by unsupported flag in __CxxFrameHandler4: 0b24:fixme:seh:__CxxFrameHandler4 unsupported flags: 7a which is unsupported FUNC_DESCR_IS_SEPARATED in dlls/vcruntime140_1/except_x86_64.c Not critical since the workaround exists. I'm ready to test this when the flag is implemented. -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.

1 12