October 2006 - Wine-devel - list.winehq.org

Request for Test: exception handling test
by Marcus Meissner 02 Oct '06

02 Oct '06

Hi, Can someone with ability to run the testsuite on Windows please test the following new 2 testcases under Windows? They succeed in WINE now, but I wonder if they will work in Windows. Ciao, Marcus --- dlls/ntdll/tests/exception.c | 145 ++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 145 insertions(+), 0 deletions(-) ef86b7cd65ade45a1e1cd269bff378166d1ad677 diff --git a/dlls/ntdll/tests/exception.c b/dlls/ntdll/tests/exception.c index 1d00816..20ff896 100644 --- a/dlls/ntdll/tests/exception.c +++ b/dlls/ntdll/tests/exception.c @@ -24,6 +24,7 @@ #define _WIN32_WINNT 0x500 /* For NTSTATUS */ #endif +#include <stdio.h> #include "ntstatus.h" #define WIN32_NO_STATUS #include "windef.h" @@ -281,6 +282,148 @@ static void test_debug_regs(void) ok(ctx.Dr7 == 0x155,"failed to set debugregister 7 to 0x155"); } +/* test the single step exception behaviour */ +static int gotsinglesteps = 0; +static DWORD single_step_handler( EXCEPTION_RECORD *rec, EXCEPTION_REGISTRATION_RECORD *frame, + CONTEXT *context, EXCEPTION_REGISTRATION_RECORD **dispatcher ) +{ + gotsinglesteps++; + ok (!(context->EFlags & 0x100), "eflags has single stepping bit set\n"); + return ExceptionContinueExecution; +} + +static BYTE single_stepcode[] = { + 0x9c, /* pushf */ + 0x58, /* pop %eax */ + 0x0d,0,1,0,0, /* or $0x100,%eax */ + 0x50, /* push %eax */ + 0x9d, /* popf */ + 0x35,0,1,0,0, /* xor $0x100,%eax */ + 0x50, /* push %eax */ + 0x9d, /* popf */ + 0xc3 +}; + +static void test_single_step(void) +{ + struct { + EXCEPTION_REGISTRATION_RECORD frame; + } exc_frame; + void (*func)(void) = (void *)single_stepcode; + + pNtCurrentTeb = (void *)GetProcAddress( GetModuleHandleA("ntdll.dll"), "NtCurrentTeb" ); + if (!pNtCurrentTeb) { + trace( "NtCurrentTeb not found, skipping tests\n" ); + return; + } + exc_frame.frame.Handler = single_step_handler; + exc_frame.frame.Prev = pNtCurrentTeb()->Tib.ExceptionList; + pNtCurrentTeb()->Tib.ExceptionList = &exc_frame.frame; + func(); + ok(gotsinglesteps == 1, "expected 1 single step exceptions, got %d\n", gotsinglesteps); + pNtCurrentTeb()->Tib.ExceptionList = exc_frame.frame.Prev; +} + +/* Test the alignment check (AC) flag handling. */ +static BYTE align_check_code[] = { + 0x55, /* push %ebp */ + 0x89,0xe5, /* mov %esp,%ebp */ + 0x9c, /* pushf */ + 0x58, /* pop %eax */ + 0x0d,0,0,4,0, /* or $0x40000,%eax */ + 0x50, /* push %eax */ + 0x9d, /* popf */ + 0x8b,0x45,8, /* mov 0x8(%ebp),%eax */ + 0xc7,0x40,1,42,0,0,0, /* movl $42,0x1(%eax) */ + 0x9c, /* pushf */ + 0x58, /* pop %eax */ + 0x35,0,0,4,0, /* xor $0x40000,%eax */ + 0x50, /* push %eax */ + 0x9d, /* popf */ + 0x5d, /* pop %ebp */ + 0xc3, /* ret */ +}; + +static int gotalignfaults = 0; +static DWORD align_check_handler( EXCEPTION_RECORD *rec, EXCEPTION_REGISTRATION_RECORD *frame, + CONTEXT *context, EXCEPTION_REGISTRATION_RECORD **dispatcher ) +{ + ok (!(context->EFlags & 0x40000), "eflags has AC bit set\n"); + gotalignfaults++; + return ExceptionContinueExecution; +} + +/* Not working in Linux now. Causes problems with unrelated functions. + * have to fix them sometime :( + */ +#undef STACK_ALIGN_TEST + +#ifdef STACK_ALIGN_TEST +static BYTE align_check_code2[] = { + 0x55, /* pushl %ebp */ + 0x89,0xe5, /* movl %esp,%ebp */ + 0x9c, /* pushf */ + 0x58, /* pop %eax */ + 0x0d,0,0,4,0, /* or $0x40000,%eax */ + 0x50, /* push %eax */ + 0x9d, /* popf */ + 0x35,0,0,4,0, /* xor $0x40000,%eax */ + 0x83,0xec,2, /* sub $0x2,%esp */ + 0x50, /* push %eax - test fault on stack */ + 0x89,0xec, /* movl %ebp, %esp */ + 0x50, /* push %eax */ + 0x9d, /* popf */ + 0x5d, /* popl %ebp */ + 0xcb /* lret */ +}; + +static DWORD align_check_handler2( EXCEPTION_RECORD *rec, EXCEPTION_REGISTRATION_RECORD *frame, + CONTEXT *context, EXCEPTION_REGISTRATION_RECORD **dispatcher ) +{ + ok (!(context->EFlags & 0x40000), "eflags has AC bit set\n"); + if (context->Esp & 0x2) { + gotalignfaults++; + context->Esp+=2; + } + return ExceptionContinueExecution; +} +#endif + + +static void test_align_faults(void) +{ + int twoints[2]; + struct { + EXCEPTION_REGISTRATION_RECORD frame; + } exc_frame; + void (*func1)(int*) = (void *)align_check_code; +#ifdef STACK_ALIGN_TEST + void (*func2)(void) = (void *)align_check_code2; +#endif + + pNtCurrentTeb = (void *)GetProcAddress( GetModuleHandleA("ntdll.dll"), "NtCurrentTeb" ); + if (!pNtCurrentTeb) { + trace( "NtCurrentTeb not found, skipping tests\n" ); + return; + } + exc_frame.frame.Handler = align_check_handler; + exc_frame.frame.Prev = pNtCurrentTeb()->Tib.ExceptionList; + pNtCurrentTeb()->Tib.ExceptionList = &exc_frame.frame; + + gotalignfaults=0; + func1(twoints); + ok(gotalignfaults == 0, "got %d alignment faults, expected 0\n", gotalignfaults); + +#ifdef STACK_ALIGN_TEST + exc_frame.frame.Handler = align_check_handler2; + gotalignfaults=0; + func2(); + ok(gotalignfaults == 0, "got %d alignment faults, expected 0\n", gotalignfaults); +#endif + + pNtCurrentTeb()->Tib.ExceptionList = exc_frame.frame.Prev; +} + #endif /* __i386__ */ START_TEST(exception) @@ -288,5 +431,7 @@ START_TEST(exception) #ifdef __i386__ test_prot_fault(); test_debug_regs(); + test_single_step(); + test_align_faults(); #endif } -- 1.2.4

1 0

no backtrace with wine crashes
by James Hawkins 02 Oct '06

02 Oct '06

Hey guys, Since very recently, any time we get a crash in Wine, winedbg does not show a backtrace. This makes it significantly more difficult to debug the crash. Is anyone else experiencing this? I'll run a regression test as soon as I can. wine: Unhandled page fault on read access to 0x00000000 at address 0x7ebfeb70 (thread 0011), starting debugger... Modules: Cannot get info on module while no process is loaded Threads: process tid prio (all id:s are in hex) 00000013 00000021 0 00000020 0 0000001e 0 0000001b 0 00000019 0 00000018 0 00000017 0 00000016 0 00000014 0 00000010 (D) (null) 00000023 0 00000022 0 00000012 0 00000011 0 0000000a 0000000c 0 0000000b 0 -- James Hawkins

2 5

Implementation question for CryptSIPLoad
by Paul Vriens 02 Oct '06

02 Oct '06

Hi, In CryptSIPLoad I have to load the some internal crypt32 function pointers into a structure. Will the following work ? : pSipDispatch->pfGet = &CryptSIPGetSignedDataMsg; pSipDispatch->pfPut = &CryptSIPPutSignedDataMsg; pSipDispatch->pfCreate = &CryptSIPCreateIndirectData; pSipDispatch->pfVerify = &CryptSIPVerifyIndirectData; pSipDispatch->pfRemove = &CryptSIPRemoveSignedDataMsg; What happens if I have another dll loaded that has the same functions (wintrust, msisip ....)? Do I need to do an explicit LoadLibrary or GetModuleHandle? The tests I've already created (in current GIT) succeed with the above 'solution' but I just want to make sure there aren't any side-effects. Cheers, Paul.

1 0

RE: compiling wine for amd64 on ubuntu dapper 6.06
by Ge van Geldorp 02 Oct '06

02 Oct '06

> From: "Gerald Britton" <gerald.britton(a)gmail.com> > > 1. /configure can't find opengl, and produces these messages: > > configure: WARNING: Wine will be build without OpenGL or > Direct3D support > configure: WARNING: because something is wrong with the OpenGL setup: > configure: WARNING: No OpenGL library found on this system. I found the easiest way around this problem is to run configure using --without-opengl. I doubt OpenGL is very useful at the moment for 64-bit Wine anyway. Ge van Geldorp.

2 1

Re: wined3d - fix grammatical errors
by Stefan Dösinger 02 Oct '06

02 Oct '06

I think Alexandre wants 1 patch per mail ;-)

1 0

Re: Copy protection
by Tim Schmidt 02 Oct '06

02 Oct '06

On 10/2/06, James Courtier-Dutton <James(a)superbug.co.uk> wrote: > The easiest way round this is to simply recognise the executable with > the copy protection, and simply install a hook to catch the appropriate > file system or registry calls and divert them to a special handling > routine to satisfy the application. The difficulty would come from > actually implementing the "copy protection" part. I.e. Preventing the > wine user from copying the software. Since we're not getting cooperation from the copy protection software houses, I don't think that's a major goal of this work. We'd like _copy protected_ software to work with Wine, who cares if the _copy protection_ software actually works beyond allowing the program to run? --tim

1 0

Full A and W implementations for registry functions in shlwapi
by Paul Vriens 02 Oct '06

02 Oct '06

Hi, is there any reason that some of the registry functions in shlwapi have a full implementation for both the A and the W calls, instead of calling W from A? (other then nobody bothered). Is it worthwhile to be done? (Could be another janitorial task). Cheers, Paul.

1 0

building outside of source tree fails - making wrong makedep
by Michael [Plouj] Ploujnikov 02 Oct '06

02 Oct '06

My Wine source is in ~/apps/wine-git. I'm trying to build Wine inside ~/apps/build_dir. So, I do this: $ mkdir ~/apps/build_dir $ cd ~/apps/build_dir $ ../wine-git/configure --x-libraries=/emul/linux/x86/usr/lib/ LDFLAGS="-L/emul/linux/x86/usr/lib -L/emul/linux/x86/lib" --verbose &> config.output # config.output attached # Then I run make, and this is where it fails: $ make make[1]: Entering directory `/home/plouj/apps/build_dir/tools' ../tools/makedep -C../../wine-git/tools -S../../wine-git -T.. -I/usr/include/fre etype2 bin2res.c fnt2bdf.c fnt2fon.c make_ctests.c makedep.c relpath.c sfnt2fnt. c make[1]: ../tools/makedep: Command not found make[1]: [Makefile] Error 127 (ignored) make[1]: `../../wine-git/tools/makedep' is up to date. make[1]: Leaving directory `/home/plouj/apps/build_dir/tools' make[1]: Entering directory `/home/plouj/apps/build_dir/libs' make[2]: Entering directory `/home/plouj/apps/build_dir/libs/port' ../../tools/makedep -C../../../wine-git/libs/port -S../../../wine-git -T../.. f fs.c fstatvfs.c futimes.c getopt.c getopt1.c getpagesize.c gettid.c interlocked. c lstat.c memcpy_unaligned.c memmove.c mkstemps.c pread.c pwrite.c readlink.c si gsetjmp.c spawn.c statvfs.c strcasecmp.c strerror.c strncasecmp.c usleep.c .... What this looks like to me is make trying to build ../../wine-git/tools/makedep which already exists, but it's trying to use ../tools/makedep, which doesn't exist, to actually build other stuff. I am trying to build Wine on Gentoo AMD64. You can see that the configure command is already convoluted to achieve that. Wine _does_ build if I run make inside the source directory. It has been working fine since before 0.9.10. That is not what I'm doing here. I don't know anything about how make works and especially about how the Makefiles in Wine are setup so I would appreciate any help to track down this problem. It's been suggested that maybe this happens because my system is broken, but I don't want to come to that conclusion without any evidence. My goal is not to simply blame someone but to actually fix what is broken. -- () ASCII Ribbon Campaign /\ - against HTML mail & vCards

2 1

Re: gdi32: Improve get_nearest_charset().
by Huw Davies 02 Oct '06

02 Oct '06

On Mon, Oct 02, 2006 at 06:42:50AM +0900, Byeong-Sik Jeon wrote: > Use the font linking information at get_nearest_charset(). > MS-Windows do it(equal behavior). Looks good to me. Thanks, Huw. -- Huw Davies huw(a)codeweavers.com

1 0

Re: wined3d - fix grammatical errors
by James Hawkins 02 Oct '06

02 Oct '06

On 10/1/06, EA Durbin <ead1234(a)hotmail.com> wrote: > > Please read up on the patch submission process [1]. Only one patch per email, and you need to include a changelog for each patch. Personally, I don't think the sentences in question need to be changed though. [1] http://winehq.org/site/sending_patches -- James Hawkins

2 2