November 2011 - Wine-Devel - list.winehq.org

Re: [2/5] cmd/tests: Standardize the batch file section headers.
by Marvin Nov. 28, 2011

Nov. 28, 2011

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=15675 Your paranoid android. === WINEBUILD (build) === Patch failed to apply

1 0

Re: kernel32/tests: Test deleting a timer right after creating it.
by Marvin Nov. 28, 2011

Nov. 28, 2011

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=15671 Your paranoid android. === W2KPROSP4 (32 bit sync) === sync.c:685: Test failed: DeleteTimerQueueTimer ret=1 le=997 === WXPPROSP3 (32 bit sync) === sync.c:685: Test failed: DeleteTimerQueueTimer ret=1 le=997 === W2K3R2SESP2 (32 bit sync) === sync.c:685: Test failed: DeleteTimerQueueTimer ret=1 le=997 === WVISTAADM (32 bit sync) === sync.c:741: Test failed: Timer callback 0 === W2K8SE (32 bit sync) === sync.c:741: Test failed: Timer callback 0 === W7PRO (32 bit sync) === sync.c:741: Test failed: Timer callback 0 === W7PROX64 (32 bit sync) === sync.c:741: Test failed: Timer callback 0 === TEST64_W7SP1 (32 bit sync) === sync.c:741: Test failed: Timer callback 0 === W7PROX64 (64 bit sync) === sync.c:741: Test failed: Timer callback 0 === TEST64_W7SP1 (64 bit sync) === sync.c:741: Test failed: Timer callback 0

1 0

Re: [PATCH] dxgi: 0 / 0 means 0 for a DXGI_RATIONAL
by Marvin Nov. 28, 2011

Nov. 28, 2011

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=15670 Your paranoid android. === WVISTAADM (32 bit swapchain) === swapchain.c:76: Test failed: Failed to create device and swapchain, hr 0x8007000e. swapchain: unhandled exception c0000005 at 004027AA === W2K8SE (32 bit swapchain) === swapchain.c:76: Test failed: Failed to create device and swapchain, hr 0x8007000e. swapchain: unhandled exception c0000005 at 004027AA === W7PRO (32 bit swapchain) === swapchain.c:76: Test failed: Failed to create device and swapchain, hr 0x8007000e. swapchain: unhandled exception c0000005 at 004027AA === W7PROX64 (32 bit swapchain) === swapchain.c:76: Test failed: Failed to create device and swapchain, hr 0x8007000e. swapchain: unhandled exception c0000005 at 004027AA === TEST64_W7SP1 (32 bit swapchain) === swapchain.c:76: Test failed: Failed to create device and swapchain, hr 0x8007000e. swapchain: unhandled exception c0000005 at 004027AA === W7PROX64 (64 bit swapchain) === swapchain.c:76: Test failed: Failed to create device and swapchain, hr 0x8007000e. Timeout === TEST64_W7SP1 (64 bit swapchain) === swapchain.c:76: Test failed: Failed to create device and swapchain, hr 0x8007000e. Timeout

1 0

Re: [PATCH 1/5] user32/tests: Initial tests for RawInput functions.
by Marvin Nov. 28, 2011

Nov. 28, 2011

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=15663 Your paranoid android. === WVISTAADM (32 bit input) === input.c:1807: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === W2K8SE (32 bit input) === input.c:1807: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === W7PRO (32 bit input) === input.c:1807: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === W7PROX64 (32 bit input) === input.c:1807: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === TEST64_W7SP1 (32 bit input) === input.c:1807: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === W7PROX64 (64 bit input) === input.c:1807: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === TEST64_W7SP1 (64 bit input) === input.c:1807: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87

2 1

Re: [PATCH 5/5] user32: Implement GetRawInputDeviceInfoW completely on user side.
by Marvin Nov. 28, 2011

Nov. 28, 2011

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=15666 Your paranoid android. === WVISTAADM (32 bit input) === input.c:1791: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === W2K8SE (32 bit input) === input.c:1791: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === W7PRO (32 bit input) === input.c:1791: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === W7PROX64 (32 bit input) === input.c:1791: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === TEST64_W7SP1 (32 bit input) === input.c:1791: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === W7PROX64 (64 bit input) === input.c:1791: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === TEST64_W7SP1 (64 bit input) === input.c:1791: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87

1 0

Re: [PATCH 3/5] user32: Implement GetRawInputDeviceList completely on user side.
by Marvin Nov. 28, 2011

Nov. 28, 2011

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=15665 Your paranoid android. === WVISTAADM (32 bit input) === input.c:1799: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === W2K8SE (32 bit input) === input.c:1799: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === W7PRO (32 bit input) === input.c:1799: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === W7PROX64 (32 bit input) === input.c:1799: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === TEST64_W7SP1 (32 bit input) === input.c:1799: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === W7PROX64 (64 bit input) === input.c:1799: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === TEST64_W7SP1 (64 bit input) === input.c:1799: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87

1 0

Re: [PATCH 2/5] user32: Add initial input parameter validations to RawInput functions.
by Marvin Nov. 28, 2011

Nov. 28, 2011

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=15664 Your paranoid android. === WVISTAADM (32 bit input) === input.c:1801: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === W2K8SE (32 bit input) === input.c:1801: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === W7PRO (32 bit input) === input.c:1801: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === W7PROX64 (32 bit input) === input.c:1801: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === TEST64_W7SP1 (32 bit input) === input.c:1801: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === W7PROX64 (64 bit input) === input.c:1801: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87 === TEST64_W7SP1 (64 bit input) === input.c:1801: Test failed: GetRawInputDeviceInfoA failed: -1, 1024, 87

1 0

Compiling Wine without prelink
by Roger Cruz Nov. 28, 2011

Nov. 28, 2011

What is the proper way to ask the configure script not to use 'prelink' on the generated shared libraries? I don't want to remove the prelink from my development platform but just tell the configurator to ignore it. Thanks Roger R. Cruz

3 5

Re: [PATCH 2/5] d3d8/tests: Test the window and backbuffer dimensions after display mode changes.
by Marvin Nov. 27, 2011

Nov. 27, 2011

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=15656 Your paranoid android. === WNT4WSSP6 (32 bit device) === Timeout

1 0

Re: [PATCH 1/5] d3d9/tests: Test the window and backbuffer dimensions after display mode changes.
by Marvin Nov. 27, 2011

Nov. 27, 2011

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=15655 Your paranoid android. === WNT4WSSP6 (32 bit device) === Timeout

1 0

Re: [PATCH] dxgi: 0 / 0 means 0 for a DXGI_RATIONAL
by Henri Verbeet Nov. 27, 2011

Nov. 27, 2011

On 27 November 2011 18:15, Eduard - Gabriel Munteanu <eduard.munteanu(a)linux360.ro> wrote: > On Sun, Nov 27, 2011 at 05:25:08PM +0100, Henri Verbeet wrote: >> What happens for cases like "60 / 0"? Do we get an exception, or does >> the function return an error? MSDN for DXGI_RATIONAL says that whole >> numbers should be represented with a denominator of 1, is something >> like "120 / 2" valid? > > I'm wasn't sure what needed to be done for stuff like 60 / 0, so I > thought that should result in division by zero for now. Maybe it should > be NaN if we convert to a float, or maybe UINT_MAX for UINT, but the > MSDN docs for DXGI_RATIONAL don't say anything about that. > > Perhaps an ERR/WARN/FIXME is in order? > > As for 120 / 2, I took "should" as a recommendation, rather than a > definitive "must". > Possibly. The point of writing the tests is to find out what actually happens.

1 0

Re: [PATCH] dxgi: 0 / 0 means 0 for a DXGI_RATIONAL
by Henri Verbeet Nov. 27, 2011

Nov. 27, 2011

On 26 November 2011 15:52, Eduard - Gabriel Munteanu <eduard.munteanu(a)linux360.ro> wrote: > + if (!rational->Denominator && !rational->Numerator) > + return 0; That's pretty much the same as returning 0 when the numerator is 0. > C_SRCS = \ > - device.c > + device.c \ > + swapchain.c > I don't think we really need a new test file for this, you can probably just add it to device.c. > + hr = D3D10CreateDeviceAndSwapChain(NULL, D3D10_DRIVER_TYPE_HARDWARE, > + NULL, 0, D3D10_SDK_VERSION, &scdesc, swapchain, device); I suppose you can test D3D10CreateDeviceAndSwapChain() for completeness if you want, but the interesting part is testing IDXGIFactory_CreateSwapChain(). What happens for cases like "60 / 0"? Do we get an exception, or does the function return an error? MSDN for DXGI_RATIONAL says that whole numbers should be represented with a denominator of 1, is something like "120 / 2" valid?

1 0

Re: [PATCH] kernel32: implement GetLogicalProcessorInformation
by Charles Davis Nov. 26, 2011

Nov. 26, 2011

Hi, On Nov 24, 2011, at 4:21 AM, Claudio Fontana wrote: > First implementation of GetLogicalProcessorInformation. > Limitations: all logical processors are added to the same NUMA set, > and all cores are added to the same package. > Only the linux-specific helper function is implemented, so for now > everybody else gets the fallback hard-coded results only. > The fallback is also triggered when the OS-specific APIs fail. > > The linux-specific function is based on sysfs. > > The new OS-specific functions can be easily added as additional > ifdefs by following the linux example. > --- > dlls/kernel32/process.c | 350 ++++++++++++++++++++++++++++++++++++++- > dlls/kernel32/tests/Makefile.in | 1 + > dlls/kernel32/tests/glpi.c | 82 +++++++++ You should really implement this in ntdll.NtQuerySystemInformation (information class SystemLogicalProcessorInformation). In general, kernel32 forwards most of its implementation to ntdll, which reduces code duplication. > 3 files changed, 430 insertions(+), 3 deletions(-) > create mode 100644 dlls/kernel32/tests/glpi.c IMO, you should add the test to dlls/ntdll/tests/info.c, and test NtQuerySystemInformation() with info class SystemLogicalProcessorInformation. > > diff --git a/dlls/kernel32/process.c b/dlls/kernel32/process.c > index b6c0b9d..de39244 100644 > --- a/dlls/kernel32/process.c > +++ b/dlls/kernel32/process.c > @@ -51,6 +51,12 @@ > > #include "ntstatus.h" > #define WIN32_NO_STATUS > + > +#include "windef.h" > +#include "winbase.h" > +#include "winerror.h" > +#include "winnt.h" > + > #include "winternl.h" > #include "kernel_private.h" > #include "psapi.h" > @@ -3705,14 +3711,352 @@ HANDLE WINAPI GetCurrentProcess(void) > return (HANDLE)~(ULONG_PTR)0; > } > > +/*************** > + * glpi_impl_fb: > + * a fallback implementation of GetLogicalProcessorInformation > + * for the systems we do not support yet, > + * or for when the OS-specific API fails. > + * Same arguments and return value as glpi. > + */ This comment should really come right before the function you're documenting. Also, it's formatted wrong. It should look more like this: /*********************************************************************************** * logical_processor_info_fallback * * Fallback implementation of NtQuerySystemInformation() with info class * SystemLogicalProcessorInfo. */ > + > +#undef SLPI > +#ifdef NONAMELESSUNION > +#define SLPI(buffer, idx) buffer[idx].DUMMYUNIONNAME > +#else > +#define SLPI(buffer, idx) buffer[idx] > +#endif /* NONAMELESSUNION */ You could avoid the ifdef if you used the U() macro. In fact, why are you defining a macro at all? Just use U(buffer[idx]). > + > +static BOOL > +glpi_impl_fb(PSYSTEM_LOGICAL_PROCESSOR_INFORMATION buffer, PDWORD pbuflen) Surely you could use a more descriptive name. (See one example in my example doc comment.) This isn't 1978, you know. > +{ > + int glpi_n = 5; > + FIXME("(%p,%p): reporting hard coded information\n", buffer, pbuflen); > + > + if (*pbuflen < sizeof(*buffer) * glpi_n) > + { > + SetLastError(ERROR_INSUFFICIENT_BUFFER); > + *pbuflen = sizeof(*buffer) * glpi_n; > + return FALSE; > + } > + > + buffer[0].ProcessorMask = (ULONG_PTR)0x01; > + buffer[1].ProcessorMask = (ULONG_PTR)0x01; > + buffer[2].ProcessorMask = (ULONG_PTR)0x01; > + buffer[3].ProcessorMask = (ULONG_PTR)0x01; > + buffer[4].ProcessorMask = (ULONG_PTR)0x01; These casts aren't necessary. > + > + buffer[0].Relationship = RelationProcessorCore; > + buffer[1].Relationship = RelationNumaNode; > + buffer[2].Relationship = RelationCache; > + buffer[3].Relationship = RelationCache; > + buffer[4].Relationship = RelationProcessorPackage; > + > + SLPI(buffer, 0).ProcessorCore.Flags = (BYTE)0x00; > + SLPI(buffer, 1).NumaNode.NodeNumber = (DWORD)0x00; Neither are these. > + > + SLPI(buffer, 2).Cache.Level = (BYTE)0x01; > + SLPI(buffer, 2).Cache.Associativity = (BYTE)0x08; > + SLPI(buffer, 2).Cache.LineSize = (WORD)64; > + SLPI(buffer, 2).Cache.Size = (DWORD)16 << 10; /* 16 KB */ Or these. > + SLPI(buffer, 2).Cache.Type = CacheData; > + > + SLPI(buffer, 3).Cache.Level = (BYTE)0x02; > + SLPI(buffer, 3).Cache.Associativity = (BYTE)0x08; > + SLPI(buffer, 3).Cache.LineSize = (WORD)64; > + SLPI(buffer, 3).Cache.Size = (DWORD)1024 << 10; /* 1024 KB */ Or these. > + SLPI(buffer, 3).Cache.Type = CacheUnified; > + /* no additional info for processor package at buffer[4] needed. */ > + > + *pbuflen = sizeof(*buffer) * glpi_n; > + return TRUE; > +} > + > +#ifdef linux > +/****************** > + * glpi_impl_linux: > + * an implementation of GetLogicalProcessorInformation > + * for linux sysfs. > + * Returns: -1 on API failure (trigger fallback), > + * otherwise see glpi. > + */ See above. > + > +#define SYS_CPU "/sys/devices/system/cpu/cpu" > +#define SYS_CPU_MAX 63 I'd criticize your use of a static limit, if only Windows didn't have that limit ;). Actually, you can only have 32 processors on a 32-bit system because a ULONG_PTR is 32 bits there. > +#define SYS_CPU_COREID "/topology/core_id" > +#define SYS_CPU_PACKAGEID "/topology/physical_package_id" > +#define SYS_CPU_CACHE "/cache/index" > +#define SYS_CPU_CACHE_MAX 12 > +#define SYS_CPU_CACHE_LEVEL "/level" > +#define SYS_CPU_CACHE_SIZE "/size" > +#define SYS_CPU_CACHE_TYPE "/type" > +#define SYS_CPU_CACHE_LINE "/coherency_line_size" > +#define SYS_CPU_CACHE_ASSOC "/ways_of_associativity" > + > +static int glpi_impl_linux(SYSTEM_LOGICAL_PROCESSOR_INFORMATION *buffer, > + PDWORD pbuflen) > +{ > + /* remember the glpi results, so that subsequent calls are faster */ > + static int n_cores = 0; > + static int n_caches = 0; > + static PSYSTEM_LOGICAL_PROCESSOR_INFORMATION cores = NULL; > + static PSYSTEM_LOGICAL_PROCESSOR_INFORMATION caches = NULL; > + static SYSTEM_LOGICAL_PROCESSOR_INFORMATION numa[1]; > + static SYSTEM_LOGICAL_PROCESSOR_INFORMATION package[1]; > + > + if (!n_cores) > + { > + int n_cpus = 0; int cpu_i; > + char path[PATH_MAX]; > + > + TRACE("building new cached result.\n"); > + > + for (n_cpus = 0; n_cpus < SYS_CPU_MAX; n_cpus++) > + { > + snprintf(path, sizeof(path), SYS_CPU "%d", n_cpus); > + if (access(path, R_OK | X_OK) != 0) Don't use access(2). There's a big fat warning on the man-page telling you not to use it because of a race condition. Another way to solve the problem you're trying to solve might be to scan the "/sys/devices/system/cpu/" directory for the number of entries using the POSIX readdir(3) interface. You could scan them all up front, or you could take each one as you go, re-allocating the table for each new entry. > + break; > + } > + > + TRACE("detected %d logical cpus.\n", n_cpus); > + if (n_cpus < 1) > + { > + TRACE("requesting fallback implementation.\n"); > + return -1; > + } > + > + cores = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, sizeof(*cores) * n_cpus); > + caches = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, sizeof(*caches) * 1); > + > + for (cpu_i = 0; cpu_i < n_cpus; cpu_i++) > + { > + FILE *f; int core_id; int i; > + snprintf(path, sizeof(path), SYS_CPU "%d" SYS_CPU_COREID, cpu_i); > + cores[n_cores].ProcessorMask = (ULONG_PTR)0x01 << cpu_i; This cast might be needed if there are more than 32 CPUs, so that's OK. > + cores[n_cores].Relationship = RelationProcessorCore; > + > + if (!(f = fopen(path, "r")) || fscanf(f, "%d", &core_id) != 1) > + { > + /* topology info not available, provide default cpuid=cpu_i */ > + ERR("could not read cpu%d topology.\n", cpu_i); > + SLPI(cores, n_cores).Reserved[0] = cpu_i; > + n_cores++; > + if (f) > + fclose(f); > + continue; /* next cpu */ > + } > + > + /* normal sane case */ > + fclose(f); > + /* store the core_id to catch cpus with same core_id */ > + SLPI(cores, n_cores).Reserved[0] = core_id; > + > + /* look for this core_id in previous elements */ > + for (i = 0; i < n_cores; i++) > + if (SLPI(cores, i).Reserved[0] == SLPI(cores, n_cores).Reserved[0]) > + break; > + > + if (i < n_cores) > + { > + TRACE("found existing coreid %d for cpu%d.\n", core_id, cpu_i); > + /* add this logical processor to the existing mask */ > + cores[i].ProcessorMask |= cores[n_cores].ProcessorMask; > + /* do not advance n_cores: this element can be overwritten. */ > + } > + else > + { > + TRACE("found new coreid %d for cpu%d.\n", core_id, cpu_i); > + > + /* look for caches */ > + for (i = 0; i < SYS_CPU_CACHE_MAX; i++) > + { > + unsigned long value; > + char string[80]; > + > + snprintf(path, sizeof(path), SYS_CPU "%d" SYS_CPU_CACHE "%d", > + cpu_i, i); > + if (access(path, R_OK | X_OK) != 0) Don't use access(2). > + break; /* no more caches found. */ > + > + caches = HeapReAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, > + caches, sizeof(*caches) * (n_caches + 1)); > + > + /* we keep the index into the cores array to set > + the right value at the end. */ > + caches[n_caches].ProcessorMask = (ULONG_PTR)n_cores; This cast, however, is definitely not needed. > + caches[n_caches].Relationship = RelationCache; > + > + snprintf(path, sizeof(path), SYS_CPU "%d" SYS_CPU_CACHE "%d" > + SYS_CPU_CACHE_LEVEL, cpu_i, i); > + if (!(f = fopen(path, "r")) || fscanf(f, "%lu", &value) != 1) > + SLPI(caches, n_caches).Cache.Level = (BYTE)0x01; Neither is this. > + else > + SLPI(caches, n_caches).Cache.Level = (BYTE)value; > + if (f) > + fclose(f); > + > + snprintf(path, sizeof(path), SYS_CPU "%d" SYS_CPU_CACHE "%d" > + SYS_CPU_CACHE_ASSOC, cpu_i, i); > + if (!(f = fopen(path, "r")) || fscanf(f, "%lu", &value) != 1) > + SLPI(caches, n_caches).Cache.Associativity = 0x08; > + else > + SLPI(caches, n_caches).Cache.Associativity = (BYTE)value; > + if (f) > + fclose(f); > + > + snprintf(path, sizeof(path), SYS_CPU "%d" SYS_CPU_CACHE "%d" > + SYS_CPU_CACHE_LINE, cpu_i, i); > + if (!(f = fopen(path, "r")) || fscanf(f, "%lu", &value) != 1) > + SLPI(caches, n_caches).Cache.LineSize = (WORD)64; This cast is not needed. > + else > + SLPI(caches, n_caches).Cache.LineSize = (WORD)value; > + if (f) > + fclose(f); > + > + snprintf(path, sizeof(path), SYS_CPU "%d" SYS_CPU_CACHE "%d" > + SYS_CPU_CACHE_SIZE, cpu_i, i); > + if (!(f = fopen(path, "r")) || fscanf(f, "%lu", &value) != 1) > + SLPI(caches, n_caches).Cache.Size = (DWORD)16 << 10; This cast isn't needed unless an int is less than 32 bits in size. > + else > + SLPI(caches, n_caches).Cache.Size = (DWORD)value << 10; > + if (f) > + fclose(f); > + > + snprintf(path, sizeof(path), SYS_CPU "%d" SYS_CPU_CACHE "%d" > + SYS_CPU_CACHE_TYPE, cpu_i, i); > + if (!(f = fopen(path, "r")) || !fgets(string, sizeof(string), f)) > + SLPI(caches, n_caches).Cache.Type = CacheData; > + else switch (string[0]) > + { > + case 'D': > + SLPI(caches, n_caches).Cache.Type = CacheData; break; > + case 'I': > + SLPI(caches, n_caches).Cache.Type = CacheInstruction; break; > + case 'U': default: > + SLPI(caches, n_caches).Cache.Type = > + string[2] == 'i' ? CacheUnified : CacheTrace; break; > + } > + > + if (f) > + fclose(f); > + > + n_caches++; > + } > + n_cores++; > + TRACE("detected %d caches for coreid %d\n", i, core_id); > + } > + } > + > + if (n_caches < 1) > + { > + HeapFree(GetProcessHeap(), 0, caches); > + caches = NULL; > + } > + > + /* trim leftovers */ > + cores = HeapReAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, > + cores, sizeof(*cores) * n_cores); > + > + if (1) /* final cleanup work */ Why the if (1)? > + { > + int i; > + /* set the real processor mask in the caches. */ > + for (i = 0; i < n_caches; i++) > + { > + int cores_i; > + cores_i = caches[i].ProcessorMask; > + caches[i].ProcessorMask = cores[cores_i].ProcessorMask; > + } > + > + /* remove the reserved values from the cores, set flags, > + build numa and package. */ > + > + numa[0].ProcessorMask = 0x00; > + numa[0].Relationship = RelationNumaNode; > + SLPI(numa, 0).NumaNode.NodeNumber = 0x00; > + package[0].Relationship = RelationProcessorPackage; > + > + for (i = 0; i < n_cores; i++) > + { > + SLPI(cores, i).Reserved[0] = 0; > + SLPI(cores, i).ProcessorCore.Flags = 0x00; /* ?unclear */ > + numa[0].ProcessorMask |= cores[i].ProcessorMask; > + } > + > + package[0].ProcessorMask = numa[0].ProcessorMask; > + } > + } > + > + TRACE("getting cached result.\n"); > + > + if (1) /* final check for required size, final result */ Same here. > + { > + DWORD required; required = sizeof(*cores) * (n_cores + n_caches + 2); > + if (*pbuflen < required) > + { > + TRACE("return with failure (ERROR_INSUFFICIENT_BUFFER).\n"); > + SetLastError(ERROR_INSUFFICIENT_BUFFER); > + *pbuflen = required; > + return 0; > + } > + > + *pbuflen = required; > + > + /* beware: buffer is SYSTEM_LOGICAL_PROCESSOR_INFORMATION *buffer; */ > + memcpy(buffer, cores, sizeof(*cores) * n_cores); > + buffer += n_cores; > + > + if (n_caches > 0) > + { > + memcpy(buffer, caches, sizeof(*caches) * n_caches); > + buffer += n_caches; > + } > + > + *buffer++ = package[0]; > + *buffer++ = numa[0]; > + } > + > + TRACE("return success!\n"); Is this really necessary? IMO it's generally better to be quiet if you have nothing interesting to say (in typical Unix fashion). And a function succeeding isn't interesting. > + return 1; > +} > + > +#undef SYS_CPU > +#undef SYS_CPU_MAX > +#undef SYS_CPU_COREID > +#undef SYS_CPU_PACKAGEID > +#undef SYS_CPU_CACHE > +#undef SYS_CPU_CACHE_MAX > +#undef SYS_CPU_CACHE_LEVEL > +#undef SYS_CPU_CACHE_SIZE > +#undef SYS_CPU_CACHE_TYPE > +#undef SYS_CPU_CACHE_LINE > +#undef SYS_CPU_CACHE_ASSOC > + > +#endif /* linux */ > + > /*********************************************************************** > * GetLogicalProcessorInformation (KERNEL32.@) > */ > BOOL WINAPI GetLogicalProcessorInformation(PSYSTEM_LOGICAL_PROCESSOR_INFORMATION buffer, PDWORD pBufLen) > { > - FIXME("(%p,%p): stub\n", buffer, pBufLen); > - SetLastError(ERROR_CALL_NOT_IMPLEMENTED); > - return FALSE; > + int rv; > + > + if (pBufLen == NULL || buffer == NULL) > + { > + SetLastError(ERROR_BAD_ARGUMENTS); > + return FALSE; > + } > + > +#if defined linux > + rv = glpi_impl_linux(buffer, pBufLen); > +#else /* no supported OS found */ > + rv = -1; > +#endif > + > + /* fallback hard-coded result */ > + if (rv < 0) > + rv = glpi_impl_fb(buffer, pBufLen); > + > + return rv; > } > > /*********************************************************************** > diff --git a/dlls/kernel32/tests/Makefile.in b/dlls/kernel32/tests/Makefile.in > index dce27db..1505944 100644 > --- a/dlls/kernel32/tests/Makefile.in > +++ b/dlls/kernel32/tests/Makefile.in > @@ -17,6 +17,7 @@ C_SRCS = \ > file.c \ > format_msg.c \ > generated.c \ > + glpi.c \ > heap.c \ > loader.c \ > locale.c \ > diff --git a/dlls/kernel32/tests/glpi.c b/dlls/kernel32/tests/glpi.c > new file mode 100644 > index 0000000..a81d1d5 > --- /dev/null > +++ b/dlls/kernel32/tests/glpi.c Like I said, this should be integrated into ntdll:info instead of being its own test in kernel32. Note that the style of reporting errors is different between ntdll and kernel32. In kernel32, the function returns a BOOL, and if it's false, you have to check the LastError. In ntdll, however, functions simply return the status code directly. Also, ntdll functions return an NTSTATUS instead of a constant from <winerror.h>. > @@ -0,0 +1,82 @@ > +#include <stdarg.h> > +#include <stdlib.h> > +#include <time.h> > +#include <stdio.h> > + > +#include "wine/test.h" > +#include "windef.h" > +#include "winbase.h" > +#include "winerror.h" > + > +static void test_glpi(void) > +{ > + SYSTEM_LOGICAL_PROCESSOR_INFORMATION buffer[200]; > + DWORD buflen = sizeof(buffer); > + BOOL ret; int i, n; > + > + ret = GetLogicalProcessorInformation(NULL, &buflen); > + ok(!ret, "Passing NULL as buffer should not be ok.\n"); > + > + ret = GetLogicalProcessorInformation(buffer, NULL); > + ok(!ret, "Passing NULL as pbuflen should not be ok.\n"); > + > + ret = GetLogicalProcessorInformation(buffer, &buflen); > + ok(ret, "Normal glpi call (%d)\n", GetLastError()); Don't call GetLastError() inside an ok(). (Actually, that's moot, because this test belongs in ntdll:process, but in general, you shouldn't call GetLastError() inside an ok().) > + > + ret = GetLogicalProcessorInformation(buffer, &buflen); > + ok(ret, "glpi call with the resulting buflen (%d)\n", buflen); Showing the last error (ntdll: return status) here might help diagnosing any problems with this test. (But don't call GetLastError() inside the ok().) > + > + buflen--; > + > + ret = GetLogicalProcessorInformation(buffer, &buflen); > + ok(!ret, "glpi call with insufficient buflen (%d)\n", buflen); Here you should make sure the LastError is ERROR_INSUFFICIENT_BUFFER (corresponding NTSTATUS: STATUS_BUFFER_TOO_SMALL, or possibly STATUS_INVALID_BUFFER_SIZE), and not some other error. > + > + ret = GetLogicalProcessorInformation(buffer, &buflen); > + ok(ret, "glpi call with resulting buflen (%d)\n", buflen); Trace the LastError (ntdll: return status) here, too. > + > + n = buflen / sizeof(SYSTEM_LOGICAL_PROCESSOR_INFORMATION); > + > + printf("glpi result: %d array elements.\n", n); Don't use printf(3); use trace(). > + > +#ifdef NONAMELESSUNION > +#define BUFFER_I buffer[i].u > +#else > +#define BUFFER_I buffer[i] > +#endif /* NONAMELESSUNION */ Like I said above, just use U(buffer[i]). > + > + for (i = 0; i < n; i++) > + { > + switch (buffer[i].Relationship) > + { > + case RelationProcessorCore: > + printf("ProcessorCore idx=%03d mask=%lxh flags=%u\n", > + i, buffer[i].ProcessorMask, BUFFER_I.ProcessorCore.Flags); > + break; > + case RelationNumaNode: > + printf("NumaNode idx=%03d mask=%lxh noden=%u\n", > + i, buffer[i].ProcessorMask, BUFFER_I.NumaNode.NodeNumber); > + break; > + case RelationCache: > + printf("Cache idx=%03d mask=%lxh " > + "l=%u, a=%u, ls=%u, sz=%u, ty=%u\n", > + i, buffer[i].ProcessorMask, > + BUFFER_I.Cache.Level, BUFFER_I.Cache.Associativity, > + BUFFER_I.Cache.LineSize, BUFFER_I.Cache.Size, > + BUFFER_I.Cache.Type); > + break; > + case RelationProcessorPackage: > + printf("ProcessorPack idx=%03d mask=%lxh\n", > + i, buffer[i].ProcessorMask); For all of these, use trace() instead of printf(3). > + break; > + default: > + ok(0, "invalid relationship %d", buffer[i].Relationship); > + } > + } > +#undef BUFFER_I > +} > + > +START_TEST(glpi) > +{ > + test_glpi(); > +} > + > -- > 1.6.4 > > > One general note about your patch: use spaces, not tabs. The rest of the file only uses tabs in doc comments. Chip

4 7

Re: dinput: Stub IDirectInputJoyConfig8 interface. (try 2)
by Jacek Caban Nov. 25, 2011

Nov. 25, 2011

Hi Vitaliy, On 11/25/11 04:17, Vitaliy Margolen wrote: > Fix QueryInterface, simplify use of AddRef, and replace "LPDIRECTINPUTJOYCONFIG8" > with "IDirectInputJoyConfig8 *". > --- > dlls/dinput/dinput_main.c | 7 ++ > dlls/dinput/dinput_private.h | 2 + > dlls/dinput/joystick.c | 196 ++++++++++++++++++++++++++++++++++ > include/dinputd.h | 243 ++++++++++++++++++++++++++++++++++++++++++ > 4 files changed, 448 insertions(+), 0 deletions(-) > create mode 100644 include/dinputd.h > + if (IsEqualGUID(&IID_IDirectInputJoyConfig8, riid )) + { + return create_joy_config( riid, ppobj ); + } This has the same problem as we discussed about msxml3 lately: http://www.winehq.org/pipermail/wine-devel/2011-November/093279.html Are you sure native is broken the same way? +static const IDirectInputJoyConfig8Vtbl JoyConfig8vt; + +HRESULT create_joy_config(REFIID riid, LPVOID *object) +{ + IDirectInputJoyConfig8Impl *new_joy_config; + + TRACE("(%s, %p)\n", debugstr_guid(riid), object); + + new_joy_config = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, sizeof(*new_joy_config)); + if (!new_joy_config) return DIERR_OUTOFMEMORY; + + new_joy_config->IDirectInputJoyConfig8_iface.lpVtbl =&JoyConfig8vt; + *object = new_joy_config; + + return DI_OK; +} If you moved create_joy_config to the bottom of the file, you wouldn't need JoyConfig8vt forward declaration. +static HRESULT WINAPI JoyConfig8Impl_QueryInterface(IDirectInputJoyConfig8 *iface, REFIID riid, void** object) +{ + IDirectInputJoyConfig8Impl *This = impl_from_IDirectInputJoyConfig8(iface); + + TRACE("(%p)->(%s, %p)\n", iface, debugstr_guid(riid), object); + + if (IsEqualGUID(&IID_IUnknown, riid) || + IsEqualGUID(&IID_IDirectInput8W, riid)) Here is a typo, it should be IID_IDirectInputJoyConfig8. Jacek

2 1

Re: [PATCH] server: assert on failure to open /dev/null (Coverity)
by Eric Pouech Nov. 25, 2011

Nov. 25, 2011

Marcus, IMO,fatal_error() should be preferred over assert() (hoping that fatal_error() is actually marked as no_return) A+ 2011/11/25 Marcus Meissner <meissner(a)suse.de> > Hi, > > if we cannot open /dev/null something is really broken. > CID 5470 > > Ciao, Marcus > --- > server/request.c | 1 + > 1 files changed, 1 insertions(+), 0 deletions(-) > > diff --git a/server/request.c b/server/request.c > index 05c7464..06fdbe8 100644 > --- a/server/request.c > +++ b/server/request.c > @@ -760,6 +760,7 @@ void open_master_socket(void) > > /* make sure the stdio fds are open */ > fd = open( "/dev/null", O_RDWR ); > + assert (fd != -1); > while (fd >= 0 && fd <= 2) fd = dup( fd ); > > if (!server_dir) fatal_error( "directory %s cannot be accessed\n", > config_dir ); > -- > 1.7.1 > > > > -- -- Eric Pouech

2 1

Re: [PATCH 1/2] kernel32: implement GetLogicalProcessorInformation
by Marvin Nov. 25, 2011

Nov. 25, 2011

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=15626 Your paranoid android. === WNT4WSSP6 (32 bit glpi) === Timeout === W2KPROSP4 (32 bit glpi) === Timeout

1 0

Re: cmd: Use WCMD_output_asis*() for strings that are not supposed to contain formating directives.
by Frédéric Delanoy Nov. 24, 2011

Nov. 24, 2011

On Thu, Nov 24, 2011 at 00:30, Francois Gouget <fgouget(a)free.fr> wrote: > Using WCMD_output*() instead could cause crashes. > --- > > For instance if a filename is '%s'. > This patch should not interfere with the previous one. > - WCMD_output(newline); > + WCMD_output_asis(newline); ... > static const WCHAR newLine2[] = {'\n','\n','\0'}; > - if (!bare) WCMD_output (newLine2); > + if (!bare) WCMD_output_asis (newLine2); ... > - WCMD_output(anykey); > + WCMD_output_asis(anykey); No need to change those, since we control those strings, and those are generic strings with no '%' directive. > - WCMD_output (version_string); > + WCMD_output_asis (version_string); This one does contain a %s ( WCMD_VERSION,"CMD Version %s\n"), but we control the version string as well (currently Wine version), so shouldn't probably be changed. > @@ -430,13 +430,13 @@ static DIRECTORY_STACK *WCMD_list_directory (DIRECTORY_STACK *inputparms, int le > strcatW(&temp[toWrite], space); > toWrite++; > if (toWrite > 99) { > - WCMD_output(temp); > + WCMD_output_asis(temp); > toWrite = 0; > strcpyW(temp, nullW); > } > padding--; > } > - WCMD_output(temp); > + WCMD_output_asis(temp); > } It's just doing padding with plain spaces, so no changes needed here. The remaining hunks should be OK. Frédéric

2 2

Re: winhttp: Resolve the server name only on the first request.
by Alexandre Julliard Nov. 24, 2011

Nov. 24, 2011

Hans Leidekker <hans(a)codeweavers.com> writes: > - send_callback( &request->hdr, WINHTTP_CALLBACK_STATUS_RESOLVING_NAME, connect->servername, strlenW(connect->servername) + 1 ); > + if (connect->sockaddr.ss_family == 0xffff) > + { This won't work on Mac OS, ss_family is a byte: gcc -m32 -c -I. -I. -I../../include -I../../include -D__WINESRC__ -D_REENTRANT -fPIC -Wall -pipe -fno-strict-aliasing -Wdeclaration-after-statement -Wempty-body -Wstrict-prototypes -Wwrite-strings -fno-omit-frame-pointer -Wpointer-arith -I/sw/include/freetype2 -I/sw/include -I/sw/include -g -O2 -o request.o request.c request.c: In function 'open_connection': request.c:931: warning: comparison is always false due to limited range of data type -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: dinput: Stub IDirectInputJoyConfig8 interface.
by Michael Stefaniuc Nov. 24, 2011

Nov. 24, 2011

Hello Vitaliy, On 11/24/2011 06:28 PM, Vitaliy Margolen wrote: > --- > dlls/dinput/dinput_main.c | 7 ++ > dlls/dinput/dinput_private.h | 2 + > dlls/dinput/joystick.c | 195 +++++++++++++++++++++++++++++++++ > include/dinputd.h | 243 ++++++++++++++++++++++++++++++++++++++++++ > 4 files changed, 447 insertions(+), 0 deletions(-) > create mode 100644 include/dinputd.h we are trying to get rid off the LPJUNK, IDirectInputJoyConfig8 * is so much more nicer than LPDIRECTINPUTJOYCONFIG8. While the above is more of a nit-pick (and an easy search and replace patch for me) the QueryInterface implementation is not correct: - Query for IID_IUnknown is not supported. - Use IDirectInputJoyConfig8_AddRef() on the interface without the cast - You can save the round trip over impl_from_IDirectInputJoyConfig8() as iface and &This->IDirectInputJoyConfig8_iface are the exact same thing. bye michael

2 2

Sugestion about wine
by Kane Jade Nov. 24, 2011

Nov. 24, 2011

Perhaps the program Zynamics BinNavi is useful to analyze the win32 applications, etc: Zynamics BinNavi is the primary binary code reverse engineering tool based on graph visualization Description BinNavi is a binary code reverse engineering tool that was built to assist vulnerability researchers who look for vulnerabilities in disassembled code. With BinNavi you can analyze disassembled x86, ARM, PowerPC, and MIPS code using the powerful built-in static code analysis techniques. In cases where static code analysis is not enough, you can use the built-in debuggers to get a live view on the program you are analyzing. Use Cases - Do *control flow based code analysis* of disassembled x86, ARM, MIPS, and PowerPC code - *Cut down on complexity* by removing unimportant code from functions - Use *powerful remote debuggers* to debug programs on different operating systems - Use Differential Debugging to *quickly locate relevant code* - Create and assign user-defined tags to *mark important functions and basic blocks* - Use *advanced code analysis algorithms* to keep track of data and code - Manage all your reverse engineering projects in *one central database* - Write *scripts and plugins to extend BinNavi* to meet your specific goals - Rename and *annotate variables and functions* to make them self-explanatory - Use the REIL meta-language to *write platform-independent program analysis code* To learn more about the features and use cases of BinNavi please take a look at the BinNavi manual<http://www.zynamics.com/binnavi/manual/index.htm> . I think that it is easier to see how they are running operations on files connected with the BinNavi program. Since is a transparent, simpler and interesting debugger. link http://www.zynamics.com/binnavi.html Forgive this text is translated by translatica.pl.

2 1

Re: [PATCH] kernel32: implement GetLogicalProcessorInformation
by Marvin Nov. 24, 2011

Nov. 24, 2011

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=15612 Your paranoid android. === WNT4WSSP6 (32 bit glpi) === Timeout === W2KPROSP4 (32 bit glpi) === Timeout

1 0

Re: cmd: Print error messages on stderr, not on stdout.
by Frédéric Delanoy Nov. 24, 2011

Nov. 24, 2011

On Thu, Nov 24, 2011 at 00:29, Francois Gouget <fgouget(a)free.fr> wrote: > --- > > I consider messages like 'Not Yet Implemented' and 'No Argument' to be > errors so that they should go to stderr. But maybe there's a strong > argument for not doing it that way. Your point is valid. I changed some/most of those myself some time ago. However the test suite fails with your patch applied (e.g. for the incomplete "for" construct). Some of the hunks might be applied though, but you'd have to run the test suite to be sure. Frédéric

1 0

Re: [PATCH] wininet: Initialize netconn_stream's content_length
by Jacek Caban Nov. 23, 2011

Nov. 23, 2011

Hi Andy, On 11/22/11 5:49 AM, Andy Clayton wrote: > The size that net_conn_read reads is limited by netconn_stream's > remaining content_length, so if the content_length is zero it reads > nothing. It also means that the read size it returns is 0, which > typically indicates that the socket was closed. As a result of the > change from 193da884ad9b982d9b740b15453aa8d3c19b054f, > HTTP_ReceiveRequestData acts on the read size of 0 and closes the > connection. HttpSendRequest calls HTTP_ReceiveRequestData if there is no > post data, closing the connection with none of the response data ever > being read. > > This patch fixes this case by initializing netconn_stream's > content_length to ~0. > > Fixes http://bugs.winehq.org/show_bug.cgi?id=27816 > This doesn't seem right. At the point where we call HTTP_ReceiveRequestData, content_length is supposed to be set based on response header. The bug must be somewhere else. Jacek

2 3

Re: po: Update French translation (try 2)
by Frédéric Delanoy Nov. 22, 2011

Nov. 22, 2011

2011/11/22 Frédéric Delanoy <frederic.delanoy(a)gmail.com>: > try 2: xB -> xO > --- > po/fr.po | 52 +++++++++++++++++++--------------------------------- > 1 files changed, 19 insertions(+), 33 deletions(-) Seems you have committed the initial version. I'll change 'bytes' to 'octets' on next po update round, therefore you may forget about this patch. Frédéric

1 0

Re: po: Update French translation
by Francois Gouget Nov. 22, 2011

Nov. 22, 2011

On Mon, 21 Nov 2011, Frédéric Delanoy wrote: [...] > msgid "%1!u!MB used, %2!u!MB available" > -msgstr "%u MB utilisés, %u MB disponibles" > +msgstr "%1!u! MB utilisés, %2!u! MB disponibles" [...] > msgctxt "unit kilobyte" > msgid "kB" > -msgstr "" > +msgstr "kB" > > #: winefile.rc:122 > msgctxt "unit megabyte" > msgid "MB" > -msgstr "" > +msgstr "MB" > > #: winefile.rc:123 > msgctxt "unit gigabyte" > msgid "GB" > -msgstr "" > +msgstr "GB" kB, MB and GB should be translated to ko, Mo and Go respectively. -- Francois Gouget <fgouget(a)free.fr> http://fgouget.free.fr/ Linux: the choice of a GNU generation

2 3