July 2019 - Wine-Devel - list.winehq.org

[PATCH v2 2/2] wineserver: Fallback to RTKIT if direct modification of thread priority failed

by Rémi Bernon

Signed-off-by: Rémi Bernon <rbernon(a)codeweavers.com> --- configure.ac | 2 +- include/config.h.in | 3 ++ server/Makefile.in | 3 +- server/thread.c | 118 +++++++++++++++++++++++++++++++++++++++++++- 4 files changed, 123 insertions(+), 3 deletions(-) diff --git a/configure.ac b/configure.ac index 4eca35b06f5..91702d4e54a 100644 --- a/configure.ac +++ b/configure.ac @@ -1386,7 +1386,7 @@ dnl **** Check for libdbus **** if test "x$with_dbus" != "xno" then WINE_PACKAGE_FLAGS(DBUS,[dbus-1],,,, - [AC_CHECK_HEADER([dbus/dbus.h], + [AC_CHECK_HEADERS([dbus/dbus.h], [WINE_CHECK_SONAME(dbus-1, dbus_connection_close,,[DBUS_CFLAGS=""],[$DBUS_LIBS])], [DBUS_CFLAGS=""])]) fi diff --git a/include/config.h.in b/include/config.h.in index 1acc02b7173..6c279f932cf 100644 --- a/include/config.h.in +++ b/include/config.h.in @@ -97,6 +97,9 @@ /* Define to 1 if you have the <curses.h> header file. */ #undef HAVE_CURSES_H +/* Define to 1 if you have the <dbus/dbus.h> header file. */ +#undef HAVE_DBUS_DBUS_H + /* Define to 1 if you have the <dirent.h> header file. */ #undef HAVE_DIRENT_H diff --git a/server/Makefile.in b/server/Makefile.in index b39bd30305b..114df2a8de3 100644 --- a/server/Makefile.in +++ b/server/Makefile.in @@ -50,4 +50,5 @@ MANPAGES = \ wineserver.fr.UTF-8.man.in \ wineserver.man.in -EXTRALIBS = $(LDEXECFLAGS) -lwine $(POLL_LIBS) $(RT_LIBS) $(INOTIFY_LIBS) +EXTRAINCL = $(DBUS_CFLAGS) +EXTRALIBS = $(LDEXECFLAGS) -lwine $(POLL_LIBS) $(RT_LIBS) $(INOTIFY_LIBS) $(DBUS_LIBS) diff --git a/server/thread.c b/server/thread.c index 1b8819bf93f..c473b18c8ea 100644 --- a/server/thread.c +++ b/server/thread.c @@ -58,6 +58,111 @@ #include "user.h" #include "security.h" +#ifdef HAVE_DBUS_DBUS_H +#include <dbus/dbus.h> + +static int rtkit_set_realtime( dbus_uint64_t process, dbus_uint64_t thread, dbus_uint32_t priority ) +{ + DBusConnection* dbus = NULL; + DBusMessage *msg = NULL, *rep = NULL; + DBusError err; + int ret = -1; + + dbus_error_init(&err); + + dbus = dbus_bus_get_private(DBUS_BUS_SYSTEM, &err); + if (dbus_error_is_set(&err)) + goto error; + + dbus_connection_set_exit_on_disconnect(dbus, 0); + + if (!(msg = dbus_message_new_method_call("org.freedesktop.RealtimeKit1", + "/org/freedesktop/RealtimeKit1", + "org.freedesktop.RealtimeKit1", + "MakeThreadRealtimeWithPID"))) + goto error; + + if (!dbus_message_append_args(msg, + DBUS_TYPE_UINT64, &process, + DBUS_TYPE_UINT64, &thread, + DBUS_TYPE_UINT32, &priority, + DBUS_TYPE_INVALID)) + goto error; + + if (!(rep = dbus_connection_send_with_reply_and_block(dbus, msg, -1, &err))) + goto error; + + if (dbus_error_is_set(&err)) + goto error; + + if (dbus_set_error_from_message(&err, rep)) + goto error; + + ret = 0; + +error: + if (rep) dbus_message_unref(rep); + if (msg) dbus_message_unref(msg); + if (dbus) + { + dbus_connection_close(dbus); + dbus_connection_unref(dbus); + } + dbus_error_free(&err); + return ret; +} + +static int rtkit_set_niceness( dbus_uint64_t process, dbus_uint64_t thread, dbus_int32_t niceness ) +{ + DBusConnection* dbus = NULL; + DBusMessage *msg = NULL, *rep = NULL; + DBusError err; + int ret = -1; + + dbus_error_init(&err); + + dbus = dbus_bus_get_private(DBUS_BUS_SYSTEM, &err); + if (dbus_error_is_set(&err)) + goto error; + + dbus_connection_set_exit_on_disconnect(dbus, 0); + + if (!(msg = dbus_message_new_method_call("org.freedesktop.RealtimeKit1", + "/org/freedesktop/RealtimeKit1", + "org.freedesktop.RealtimeKit1", + "MakeThreadHighPriorityWithPID"))) + goto error; + + if (!dbus_message_append_args(msg, + DBUS_TYPE_UINT64, &process, + DBUS_TYPE_UINT64, &thread, + DBUS_TYPE_INT32, &niceness, + DBUS_TYPE_INVALID)) + goto error; + + if (!(rep = dbus_connection_send_with_reply_and_block(dbus, msg, -1, &err))) + goto error; + + if (dbus_error_is_set(&err)) + goto error; + + if (dbus_set_error_from_message(&err, rep)) + goto error; + + ret = 0; + +error: + if (rep) dbus_message_unref(rep); + if (msg) dbus_message_unref(msg); + if (dbus) + { + dbus_connection_close(dbus); + dbus_connection_unref(dbus); + } + dbus_error_free(&err); + return ret; +} +#endif #ifdef __i386__ static const unsigned int supported_cpus = CPU_FLAG(CPU_x86); @@ -522,7 +627,8 @@ affinity_t get_thread_affinity( struct thread *thread ) return mask; } -#if defined(HAVE_SCHED_SETSCHEDULER) || defined(HAVE_SETPRIORITY) +#if defined(HAVE_SCHED_SETSCHEDULER) || defined(HAVE_SETPRIORITY) || \ + defined(HAVE_DBUS_DBUS_H) static int get_unix_priority( int priority_class, int priority ) { switch (priority_class) { @@ -641,6 +747,11 @@ int set_thread_priority( struct thread* thread, int priority_class, int priority param.sched_priority = get_unix_priority( priority_class, priority ); if (sched_setscheduler( thread->unix_tid, SCHED_RR|SCHED_RESET_ON_FORK, &param ) == 0) return 0; +#endif +#ifdef HAVE_DBUS_DBUS_H + if (rtkit_set_realtime( thread->unix_pid, thread->unix_tid, + get_unix_priority( priority_class, priority ) ) == 0) + return 0; #endif } else @@ -649,6 +760,11 @@ int set_thread_priority( struct thread* thread, int priority_class, int priority if (setpriority( PRIO_PROCESS, thread->unix_tid, get_unix_priority( priority_class, priority ) ) == 0) return 0; +#endif +#ifdef HAVE_DBUS_DBUS_H + if (rtkit_set_niceness( thread->unix_pid, thread->unix_tid, + get_unix_priority( priority_class, priority ) ) == 0) + return 0; #endif } -- 2.20.1

July 8, 2019

[PATCH v2 1/2] wineserver: Implement thread priorities with sched_setscheduler / setpriority

by Rémi Bernon

This does not report permission errors in order to avoid any breaking change, only the parameter checks that were already there are returning errors. Signed-off-by: Rémi Bernon <rbernon(a)codeweavers.com> --- configure.ac | 19 ++++++ include/config.h.in | 6 ++ server/process.c | 20 +++++- server/thread.c | 152 ++++++++++++++++++++++++++++++++++++++++---- server/thread.h | 1 + 5 files changed, 184 insertions(+), 14 deletions(-) diff --git a/configure.ac b/configure.ac index a7c45ace738..4eca35b06f5 100644 --- a/configure.ac +++ b/configure.ac @@ -2234,6 +2234,25 @@ then AC_DEFINE(HAVE_SCHED_SETAFFINITY, 1, [Define to 1 if you have the `sched_setaffinity' function.]) fi +AC_CACHE_CHECK([for sched_setscheduler],wine_cv_have_sched_setscheduler, + AC_LINK_IFELSE([AC_LANG_PROGRAM( +[[#define _GNU_SOURCE +#include <sched.h>]], [[sched_setscheduler(0, 0, 0);]])],[wine_cv_have_sched_setscheduler=yes],[wine_cv_have_sched_setscheduler=no])) +if test "$wine_cv_have_sched_setscheduler" = "yes" +then + AC_DEFINE(HAVE_SCHED_SETSCHEDULER, 1, [Define to 1 if you have the `sched_setscheduler' function.]) +fi + +AC_CACHE_CHECK([for setpriority],wine_cv_have_setpriority, + AC_LINK_IFELSE([AC_LANG_PROGRAM( +[[#define _GNU_SOURCE +#include <sys/resource.h> +#include <sys/time.h>]], [[setpriority(0, 0, 0);]])],[wine_cv_have_setpriority=yes],[wine_cv_have_setpriority=no])) +if test "$wine_cv_have_setpriority" = "yes" +then + AC_DEFINE(HAVE_SETPRIORITY, 1, [Define to 1 if you have the `setpriority' function.]) +fi + AC_CACHE_CHECK([for fallocate],wine_cv_have_fallocate, AC_LINK_IFELSE([AC_LANG_PROGRAM( [[#define _GNU_SOURCE diff --git a/include/config.h.in b/include/config.h.in index ce5c1570541..1acc02b7173 100644 --- a/include/config.h.in +++ b/include/config.h.in @@ -788,6 +788,9 @@ /* Define to 1 if you have the `sched_setaffinity' function. */ #undef HAVE_SCHED_SETAFFINITY +/* Define to 1 if you have the `sched_setscheduler' function. */ +#undef HAVE_SCHED_SETSCHEDULER + /* Define to 1 if you have the `sched_yield' function. */ #undef HAVE_SCHED_YIELD @@ -812,6 +815,9 @@ /* Define to 1 if you have the `select' function. */ #undef HAVE_SELECT +/* Define to 1 if you have the `setpriority' function. */ +#undef HAVE_SETPRIORITY + /* Define to 1 if you have the `setproctitle' function. */ #undef HAVE_SETPROCTITLE diff --git a/server/process.c b/server/process.c index b67bd88da47..bf216a76f78 100644 --- a/server/process.c +++ b/server/process.c @@ -1464,6 +1464,24 @@ DECL_HANDLER(get_process_vm_counters) release_object( process ); } +static void set_process_priority( struct process *process, int priority ) +{ + struct thread *thread; + + if (!process->running_threads) + { + set_error( STATUS_PROCESS_IS_TERMINATING ); + return; + } + + LIST_FOR_EACH_ENTRY( thread, &process->thread_list, struct thread, proc_entry ) + { + set_thread_priority( thread, priority, thread->priority ); + } + + process->priority = priority; +} + static void set_process_affinity( struct process *process, affinity_t affinity ) { struct thread *thread; @@ -1489,7 +1507,7 @@ DECL_HANDLER(set_process_info) if ((process = get_process_from_handle( req->handle, PROCESS_SET_INFORMATION ))) { - if (req->mask & SET_PROCESS_INFO_PRIORITY) process->priority = req->priority; + if (req->mask & SET_PROCESS_INFO_PRIORITY) set_process_priority( process, req->priority ); if (req->mask & SET_PROCESS_INFO_AFFINITY) set_process_affinity( process, req->affinity ); release_object( process ); } diff --git a/server/thread.c b/server/thread.c index d5742a1bf96..1b8819bf93f 100644 --- a/server/thread.c +++ b/server/thread.c @@ -38,6 +38,12 @@ #ifdef HAVE_SCHED_H #include <sched.h> #endif +#ifdef HAVE_SYS_TIME_H +#include <sys/time.h> +#endif +#ifdef HAVE_SYS_RESOURCE_H +#include <sys/resource.h> +#endif #include "ntstatus.h" #define WIN32_NO_STATUS @@ -516,28 +522,147 @@ affinity_t get_thread_affinity( struct thread *thread ) return mask; } +#if defined(HAVE_SCHED_SETSCHEDULER) || defined(HAVE_SETPRIORITY) +static int get_unix_priority( int priority_class, int priority ) +{ + switch (priority_class) { + case PROCESS_PRIOCLASS_IDLE: + switch (priority) { + case THREAD_PRIORITY_IDLE: return 15; + case THREAD_PRIORITY_LOWEST: return 10; + case THREAD_PRIORITY_BELOW_NORMAL: return 8; + case THREAD_PRIORITY_NORMAL: return 6; + case THREAD_PRIORITY_ABOVE_NORMAL: return 4; + case THREAD_PRIORITY_HIGHEST: return 2; + case THREAD_PRIORITY_TIME_CRITICAL: return -15; + } + case PROCESS_PRIOCLASS_BELOW_NORMAL: + switch (priority) { + case THREAD_PRIORITY_IDLE: return 15; + case THREAD_PRIORITY_LOWEST: return 8; + case THREAD_PRIORITY_BELOW_NORMAL: return 6; + case THREAD_PRIORITY_NORMAL: return 4; + case THREAD_PRIORITY_ABOVE_NORMAL: return 2; + case THREAD_PRIORITY_HIGHEST: return 0; + case THREAD_PRIORITY_TIME_CRITICAL: return -15; + } + case PROCESS_PRIOCLASS_NORMAL: + switch (priority) { + case THREAD_PRIORITY_IDLE: return 15; + case THREAD_PRIORITY_LOWEST: return 4; + case THREAD_PRIORITY_BELOW_NORMAL: return 2; + case THREAD_PRIORITY_NORMAL: return 0; + case THREAD_PRIORITY_ABOVE_NORMAL: return -2; + case THREAD_PRIORITY_HIGHEST: return -4; + case THREAD_PRIORITY_TIME_CRITICAL: return -15; + } + case PROCESS_PRIOCLASS_ABOVE_NORMAL: + switch (priority) { + case THREAD_PRIORITY_IDLE: return 15; + case THREAD_PRIORITY_LOWEST: return 0; + case THREAD_PRIORITY_BELOW_NORMAL: return -2; + case THREAD_PRIORITY_NORMAL: return -4; + case THREAD_PRIORITY_ABOVE_NORMAL: return -6; + case THREAD_PRIORITY_HIGHEST: return -8; + case THREAD_PRIORITY_TIME_CRITICAL: return -15; + } + case PROCESS_PRIOCLASS_HIGH: + switch (priority) { + case THREAD_PRIORITY_IDLE: return 15; + case THREAD_PRIORITY_LOWEST: return -2; + case THREAD_PRIORITY_BELOW_NORMAL: return -4; + case THREAD_PRIORITY_NORMAL: return -6; + case THREAD_PRIORITY_ABOVE_NORMAL: return -8; + case THREAD_PRIORITY_HIGHEST: return -10; + case THREAD_PRIORITY_TIME_CRITICAL: return -15; + } + case PROCESS_PRIOCLASS_REALTIME: + switch (priority) { + case THREAD_PRIORITY_IDLE: return 1; + case -7: + case -6: + case -5: + case -4: + case -3: + case THREAD_PRIORITY_LOWEST: + case THREAD_PRIORITY_BELOW_NORMAL: + case THREAD_PRIORITY_NORMAL: + case THREAD_PRIORITY_ABOVE_NORMAL: + case THREAD_PRIORITY_HIGHEST: + case 3: + case 4: + case 5: + case 6: + return priority + 9; + case THREAD_PRIORITY_TIME_CRITICAL: + return 16; + } + } + return 0; +} +#endif + #define THREAD_PRIORITY_REALTIME_HIGHEST 6 #define THREAD_PRIORITY_REALTIME_LOWEST -7 +int set_thread_priority( struct thread* thread, int priority_class, int priority ) +{ + int max = THREAD_PRIORITY_HIGHEST; + int min = THREAD_PRIORITY_LOWEST; + if (priority_class == PROCESS_PRIOCLASS_REALTIME) + { + max = THREAD_PRIORITY_REALTIME_HIGHEST; + min = THREAD_PRIORITY_REALTIME_LOWEST; + } + + if ((priority < min || priority > max) && + priority != THREAD_PRIORITY_IDLE && + priority != THREAD_PRIORITY_TIME_CRITICAL) + { + errno = EINVAL; + return -1; + } + + if (thread->process->priority == priority_class && + thread->priority == priority) + return 0; + + thread->priority = priority; + if (thread->unix_tid == -1) + return 0; + + if (priority_class == PROCESS_PRIOCLASS_REALTIME) + { +#ifdef HAVE_SCHED_SETSCHEDULER + struct sched_param param; + if (sched_getparam( thread->unix_tid, &param ) != 0) + return 0; /* ignore errors for now */ + + param.sched_priority = get_unix_priority( priority_class, priority ); + if (sched_setscheduler( thread->unix_tid, SCHED_RR|SCHED_RESET_ON_FORK, &param ) == 0) + return 0; +#endif + } + else + { +#ifdef HAVE_SETPRIORITY + if (setpriority( PRIO_PROCESS, thread->unix_tid, + get_unix_priority( priority_class, priority ) ) == 0) + return 0; +#endif + } + + return 0; /* ignore errors for now */ +} + /* set all information about a thread */ static void set_thread_info( struct thread *thread, const struct set_thread_info_request *req ) { if (req->mask & SET_THREAD_INFO_PRIORITY) { - int max = THREAD_PRIORITY_HIGHEST; - int min = THREAD_PRIORITY_LOWEST; - if (thread->process->priority == PROCESS_PRIOCLASS_REALTIME) - { - max = THREAD_PRIORITY_REALTIME_HIGHEST; - min = THREAD_PRIORITY_REALTIME_LOWEST; - } - if ((req->priority >= min && req->priority <= max) || - req->priority == THREAD_PRIORITY_IDLE || - req->priority == THREAD_PRIORITY_TIME_CRITICAL) - thread->priority = req->priority; - else - set_error( STATUS_INVALID_PARAMETER ); + if (set_thread_priority( thread, thread->process->priority, req->priority )) + file_set_error(); } if (req->mask & SET_THREAD_INFO_AFFINITY) { @@ -1368,6 +1493,7 @@ DECL_HANDLER(init_thread) process->unix_pid = -1; /* can happen with linuxthreads */ init_thread_context( current ); generate_debug_event( current, CREATE_THREAD_DEBUG_EVENT, &req->entry ); + set_thread_priority( current, current->process->priority, current->priority ); set_thread_affinity( current, current->affinity ); } debug_level = max( debug_level, req->debug_level ); diff --git a/server/thread.h b/server/thread.h index 40f0eecbf1b..5e58a1bc683 100644 --- a/server/thread.h +++ b/server/thread.h @@ -126,6 +126,7 @@ extern int thread_add_inflight_fd( struct thread *thread, int client, int server extern int thread_get_inflight_fd( struct thread *thread, int client ); extern struct thread_snapshot *thread_snap( int *count ); extern struct token *thread_get_impersonation_token( struct thread *thread ); +extern int set_thread_priority( struct thread *thread, int priority_class, int priority ); extern int set_thread_affinity( struct thread *thread, affinity_t affinity ); extern int is_cpu_supported( enum cpu_type cpu ); extern unsigned int get_supported_cpu_mask(void); -- 2.20.1

July 8, 2019

Re: [PATCH 1/2] dxgi/tests: Test output after using Alt+Enter to switch fullscreen.

by Henri Verbeet

On Mon, 8 Jul 2019 at 11:39, Zhiyi Zhang <zzhang(a)codeweavers.com> wrote: > @@ -5448,11 +5449,15 @@ static void test_window_association(void) > PostMessageA(swapchain_desc.OutputWindow, WM_SYSKEYDOWN, VK_RETURN, > (MapVirtualKeyA(VK_RETURN, MAPVK_VK_TO_VSC) << 16) | 0x20000001); > flush_events(); > - hr = IDXGISwapChain_GetFullscreenState(swapchain, &fullscreen, NULL); > + output = NULL; > + hr = IDXGISwapChain_GetFullscreenState(swapchain, &fullscreen, &output); > ok(hr == S_OK, "Test %u: Got unexpected hr %#x.\n", i, hr); > ok(fullscreen == tests[i].expect_fullscreen > || broken(tests[i].broken_d3d10 && fullscreen), > "Test %u: Got unexpected fullscreen %#x.\n", i, fullscreen); > + todo_wine_if(fullscreen) ok(fullscreen ? !!output : !output, "Got wrong output.\n"); Since this is run inside a loop, you can't tell from the line number alone which test failed. > wndproc = GetWindowLongPtrW(swapchain_desc.OutputWindow, GWLP_WNDPROC); > ok(wndproc == original_wndproc, "Text %u: Got unexpected wndproc %#lx, expected %#lx.\n", Incidentally, I see I made a typo ("Text") above.

July 8, 2019

Re: [PATCH] wbemprox: Return WBEM_E_ACCESS_DENIED from IEnumWbemClassObject::Next() if the object is empty.

by Hans Leidekker

On Mon, 2019-07-08 at 20:56 +0800, Dmitry Timoshkov wrote: > Hans Leidekker <hans(a)codeweavers.com> wrote: > > > On Mon, 2019-07-08 at 20:09 +0800, Dmitry Timoshkov wrote: > > > Dmitry Timoshkov <dmitry(a)baikal.ru> wrote: > > > > > > > > > Could you suggest a place in Wine where a permission check could be added? > > > > > > > > > > It's not clear to me if this is a general WMI thing or if it depends on the > > > > > object being queried. > > > > > > > > I'll copy WMI tests from dlls/wbemprox/tests to the MSI custom action tests > > > > to get a picture how it works under Windows. > > > > > > I've ported dlls/wbemprox/query.c to run from the MSI custom action, > > > added strict checks for ::Next() return values everywhere, and while > > > some of the queries fail ::Next() never returns WBEM_E_ACCESS_DENIED. > > > It turned out that commenting out CoSetProxyBlanket() made the ::Next() > > > calls for empty objects always return WBEM_E_ACCESS_DENIED instead of > > > WBEM_S_FALSE/WBEM_S_TIMEDOUT. It's not obvious to me where the proposed > > > access check should be added. Hans, do the test results add any clue to you? > > > > Commenting out the CoSetProxyBlanket call in the wbemprox tests doesn't change > > anything however. > > Yes, sorry for not mentioning that, since I'm testing on both sides. > > > We still don't know what's different when running as a custom > > action. > > What exactly you would like to know? Also if you have suggestions for > further tests and things to examine that would help a lot. Check the process token, for example, and compare properties/privileges.

July 8, 2019

Re: [PATCH] wbemprox: Return WBEM_E_ACCESS_DENIED from IEnumWbemClassObject::Next() if the object is empty.

by Dmitry Timoshkov

Hans Leidekker <hans(a)codeweavers.com> wrote: > On Mon, 2019-07-08 at 20:09 +0800, Dmitry Timoshkov wrote: > > Dmitry Timoshkov <dmitry(a)baikal.ru> wrote: > > > > > > > Could you suggest a place in Wine where a permission check could be added? > > > > > > > > It's not clear to me if this is a general WMI thing or if it depends on the > > > > object being queried. > > > > > > I'll copy WMI tests from dlls/wbemprox/tests to the MSI custom action tests > > > to get a picture how it works under Windows. > > > > I've ported dlls/wbemprox/query.c to run from the MSI custom action, > > added strict checks for ::Next() return values everywhere, and while > > some of the queries fail ::Next() never returns WBEM_E_ACCESS_DENIED. > > It turned out that commenting out CoSetProxyBlanket() made the ::Next() > > calls for empty objects always return WBEM_E_ACCESS_DENIED instead of > > WBEM_S_FALSE/WBEM_S_TIMEDOUT. It's not obvious to me where the proposed > > access check should be added. Hans, do the test results add any clue to you? > > Commenting out the CoSetProxyBlanket call in the wbemprox tests doesn't change > anything however. Yes, sorry for not mentioning that, since I'm testing on both sides. > We still don't know what's different when running as a custom > action. What exactly you would like to know? Also if you have suggestions for further tests and things to examine that would help a lot. -- Dmitry.

July 8, 2019

Re: [PATCH] wbemprox: Return WBEM_E_ACCESS_DENIED from IEnumWbemClassObject::Next() if the object is empty.

by Hans Leidekker

On Mon, 2019-07-08 at 20:09 +0800, Dmitry Timoshkov wrote: > Dmitry Timoshkov <dmitry(a)baikal.ru> wrote: > > > > > Could you suggest a place in Wine where a permission check could be added? > > > > > > It's not clear to me if this is a general WMI thing or if it depends on the > > > object being queried. > > > > I'll copy WMI tests from dlls/wbemprox/tests to the MSI custom action tests > > to get a picture how it works under Windows. > > I've ported dlls/wbemprox/query.c to run from the MSI custom action, > added strict checks for ::Next() return values everywhere, and while > some of the queries fail ::Next() never returns WBEM_E_ACCESS_DENIED. > It turned out that commenting out CoSetProxyBlanket() made the ::Next() > calls for empty objects always return WBEM_E_ACCESS_DENIED instead of > WBEM_S_FALSE/WBEM_S_TIMEDOUT. It's not obvious to me where the proposed > access check should be added. Hans, do the test results add any clue to you? Commenting out the CoSetProxyBlanket call in the wbemprox tests doesn't change anything however. We still don't know what's different when running as a custom action.

July 8, 2019

Re: [PATCH] wbemprox: Return WBEM_E_ACCESS_DENIED from IEnumWbemClassObject::Next() if the object is empty.

by Dmitry Timoshkov

Dmitry Timoshkov <dmitry(a)baikal.ru> wrote: > > > Could you suggest a place in Wine where a permission check could be added? > > > > It's not clear to me if this is a general WMI thing or if it depends on the > > object being queried. > > I'll copy WMI tests from dlls/wbemprox/tests to the MSI custom action tests > to get a picture how it works under Windows. I've ported dlls/wbemprox/query.c to run from the MSI custom action, added strict checks for ::Next() return values everywhere, and while some of the queries fail ::Next() never returns WBEM_E_ACCESS_DENIED. It turned out that commenting out CoSetProxyBlanket() made the ::Next() calls for empty objects always return WBEM_E_ACCESS_DENIED instead of WBEM_S_FALSE/WBEM_S_TIMEDOUT. It's not obvious to me where the proposed access check should be added. Hans, do the test results add any clue to you? -- Dmitry.

July 8, 2019

[website] German translation for release 4.12.1

by Julian Rüger

Signed-off-by: Julian Rüger <jr98(a)gmx.net>

July 8, 2019

[website] German translation for release 4.12

by Julian Rüger

Signed-off-by: Julian Rüger <jr98(a)gmx.net>

July 8, 2019

Re: [PATCH] wbemprox: Return WBEM_E_ACCESS_DENIED from IEnumWbemClassObject::Next() if the object is empty.

by Dmitry Timoshkov

Hans Leidekker <hans(a)codeweavers.com> wrote: > On Mon, 2019-07-08 at 17:37 +0800, Dmitry Timoshkov wrote: > > Looks like this is a result of executing from an MSI custom action. > > > > Is there a way to detect such a case and return an appropriate error > > > > value? > > > > > > The error code suggests that the custom action is run with insufficient > > > permissions. If that's the case we'd need to do the same and add a permission > > > check somewhere. > > > > Under Windows MSI custom action is executed in the context of an MSI server, > > which probably runs as a restricted process, and which doesn't exist in Wine. > > Custom actions are executed in a seperate process in Wine. Yes, but that's obviously not an MSI server which probably executes under a resticted user account in Windows. > > Could you suggest a place in Wine where a permission check could be added? > > It's not clear to me if this is a general WMI thing or if it depends on the > object being queried. I'll copy WMI tests from dlls/wbemprox/tests to the MSI custom action tests to get a picture how it works under Windows. -- Dmitry.

July 8, 2019