xcb issues on update from 6.0.0 to 6.2.0-beta3

ltmdum

@JKSH Thanks. I have a no for EGL on X11 but I also had that for 6.0.0, so I thought it was fine. Is that my problem?
Here is my summary

Building for: linux-g++ (x86_64, CPU features: mmx sse sse2)
Compiler: gcc 9.1.0
Build options:
  Mode ................................... debug
  Optimize debug build ................... yes
  Fully optimize release builds (-O3) .... no
  Building shared libraries .............. no
  Using C standard ....................... C11
  Using C++ standard ..................... C++14
  Using ccache ........................... no
  Using new DTAGS ........................ yes
  Generating GDB index ................... no
  Relocatable ............................ no
  Using precompiled headers .............. yes
  Using LTCG ............................. no
  Target compiler supports:
    SSE .................................. SSE2 SSE3 SSSE3 SSE4.1 SSE4.2
    AVX .................................. AVX AVX2
    AVX512 ............................... F ER CD PF DQ BW VL IFMA VBMI
    Other x86 ............................ AES F16C RDRAND SHA
    Intrinsics without compiler architecture option  yes
  Sanitizers:
    Addresses ............................ no
    Threads .............................. no
    Memory ............................... no
    Fuzzer (instrumentation only) ........ no
    Undefined ............................ no
  Build parts ............................ libs tools
Qt modules and options:
  Qt Concurrent .......................... yes
  Qt D-Bus ............................... yes
  Qt D-Bus directly linked to libdbus .... no
  Qt Gui ................................. yes
  Qt Network ............................. yes
  Qt Testlib ............................. yes
  Qt Widgets ............................. yes
  Qt Xml ................................. yes
Support enabled for:
  Using pkg-config ....................... yes
  udev ................................... no
  Using system zlib ...................... yes
  Zstandard support ...................... no
  Thread support ......................... yes
Common build options:
  Linker can resolve circular dependencies  yes
Qt Core:
  backtrace .............................. yes
  DoubleConversion ....................... yes
    Using system DoubleConversion ........ no
  GLib ................................... yes
  ICU .................................... no
  Using system libb2 ..................... no
  Built-in copy of the MIME database ..... yes
  Tracing backend ........................ <none>
  Logging backends:
    journald ............................. no
    syslog ............................... no
    slog2 ................................ no
  PCRE2 .................................. yes
    Using system PCRE2 ................... no
  CLONE_PIDFD support in forkfd .......... yes
Qt Network:
  getifaddrs() ........................... yes
  IPv6 ifname ............................ yes
  libproxy ............................... no
  Linux AF_NETLINK ....................... no
  OpenSSL ................................ no
    Qt directly linked to OpenSSL ........ no
  OpenSSL 1.1 ............................ no
  DTLS ................................... no
  OCSP-stapling .......................... no
  SCTP ................................... no
  Use system proxies ..................... yes
  GSSAPI ................................. yes
  Brotli Decompression Support ........... no
Qt Gui:
  Accessibility .......................... yes
  FreeType ............................... yes
    Using system FreeType ................ yes
  HarfBuzz ............................... yes
    Using system HarfBuzz ................ no
  Fontconfig ............................. yes
  Image formats:
    GIF .................................. yes
    ICO .................................. yes
    JPEG ................................. yes
      Using system libjpeg ............... no
    PNG .................................. yes
      Using system libpng ................ yes
  Text formats:
    HtmlParser ........................... yes
    CssParser ............................ yes
    OdfWriter ............................ yes
    MarkdownReader ....................... yes
      Using system libmd4c ............... no
    MarkdownWriter ....................... yes
  EGL .................................... no
  OpenVG ................................. no
  OpenGL:
    Desktop OpenGL ....................... yes
    OpenGL ES 2.0 ........................ no
    OpenGL ES 3.0 ........................ no
    OpenGL ES 3.1 ........................ no
    OpenGL ES 3.2 ........................ no
  Vulkan ................................. no
  Session Management ..................... yes
Features used by QPA backends:
  evdev .................................. yes
  libinput ............................... no
  INTEGRITY HID .......................... no
  mtdev .................................. no
  tslib .................................. no
  xkbcommon .............................. yes
  X11 specific:
    XLib ................................. yes
    XCB Xlib ............................. yes
    EGL on X11 ........................... no
    xkbcommon-x11 ........................ yes
    xcb-sm ............................... yes
QPA backends:
  DirectFB ............................... no
  EGLFS .................................. no
  EGLFS details:
    EGLFS OpenWFD ........................ no
    EGLFS i.Mx6 .......................... no
    EGLFS i.Mx6 Wayland .................. no
    EGLFS RCAR ........................... no
    EGLFS EGLDevice ...................... no
    EGLFS GBM ............................ no
    EGLFS VSP2 ........................... no
    EGLFS Mali ........................... no
    EGLFS Raspberry Pi ................... no
    EGLFS X11 ............................ no
  LinuxFB ................................ yes
  VNC .................................... yes
  VK_KHR_display ......................... no
  QNX:
    lgmon ................................ no
    IMF .................................. no
  XCB:
    Using system-provided xcb-xinput ..... yes
    GL integrations:
      GLX Plugin ......................... yes
        XCB GLX .......................... yes
      EGL-X11 Plugin ..................... no
  Windows:
    Direct 2D ............................ no
    Direct 2D 1.1 ........................ no
    DirectWrite .......................... no
    DirectWrite 3 ........................ no
Qt Widgets:
  GTK+ ................................... no
  Styles ................................. Fusion Windows
Qt Testlib:
  Tester for item models ................. yes
Qt PrintSupport:
  CUPS ................................... no
Core tools:
  qmake tool ............................. yes

SGaist

Does it also happen if you run a shared build ?

JKSH

@ltmdum said in xcb issues on update from 6.0.0 to 6.2.0-beta3:

I have a no for EGL on X11 but I also had that for 6.0.0, so I thought it was fine. Is that my problem?

Hmm... good question; I'm not sure. If it worked in 6.0.0, I'd expect it to work with 6.2.0 too.

Are you able to compare the configure.summary files for 6.0 and 6.2? Are there any differences at all?

ltmdum

@SGaist a shared build is a last resort. I know you're asking for testing purposes but it will take some work for me to create one.

@JKSH I had a look at a diff between the two files and the difference was negligible.

I am beginning to think that the issue lies in cmake/QtPluginHelpers.cmake. I have found that with 6.0.0 the generated cpp files with the Q_IMPORT_PLUGIN macro are created where my final binary is but the equivalent files in 6.2.0-beta3 go to src/plugins/... directory. I have since found that my 6.0.0 binary incorporates these files but the 6.2.0-beta3 binary does not. There have been a lot of changes around this area of the source between 6.0.0 and 6.2.0 and the way things are done is completely different. Still trying to get my head around it

SGaist

@ltmdum said in xcb issues on update from 6.0.0 to 6.2.0-beta3:

@SGaist a shared build is a last resort. I know you're asking for testing purposes but it will take some work for me to create one

Maybe less than you think, build only qtbase with the same configuration you have now just dropping the static option. It will be faster.

ltmdum

@SGaist You're right, that would have been faster. However, I have figured out my problem... We have QT_NO_CREATE_TARGETS set to True. There is a reason for that because setting it to false breaks my build but that is a new problem.

@SGaist @JKSH thanks for your help

JKSH

@ltmdum Glad to hear that you've tracked it down!

ltmdum

I'm finally up and running, just as an FYI I had to make these changes to firstly get past a "target_sources can not be used on an ALIAS target" error and secondly to get past "cmake objects of target <blah> referenced but no such target exists"

diff --git a/cmake/QtPublicPluginHelpers.cmake b/cmake/QtPublicPluginHelpers.cmake
index 5023498b..1750cf63 100644
--- a/cmake/QtPublicPluginHelpers.cmake
+++ b/cmake/QtPublicPluginHelpers.cmake
@@ -193,12 +193,18 @@ function(__qt_internal_add_static_plugin_import_macro
         qt_module_unprefixed)

     __qt_internal_get_static_plugin_init_target_name("${plugin_target}" plugin_init_target)
-    set(objs_genex "$<TARGET_OBJECTS:${QT_CMAKE_EXPORT_NAMESPACE}::${plugin_init_target}>")
+    set(objs_genex "$<TARGET_OBJECTS:${plugin_init_target}>")
+
+    set(_plugin_target "${plugin_target}")
+    get_target_property(__aliased_target ${_plugin_target} ALIASED_TARGET)
+    if(__aliased_target)
+        set(_plugin_target ${__aliased_target})
+    endif()

     # TODO: Replace the target_sources call here with the more generalized approach that will be
     # added to __qt_propagate_generated_resource (the one that determines whether to use
     # target_sources or target_link_libraries depending on the CMake version).
-    target_sources(${QT_CMAKE_EXPORT_NAMESPACE}::${plugin_target} INTERFACE
+    target_sources(${_plugin_target} INTERFACE
                    "${objs_genex}")
 endfunction()

SGaist

Did you consider submitting this patch for inclusion ?

ltmdum

@SGaist I hadn't but now I am. I will look into it. I think my patch could be cleaner

ltmdum

@ltmdum hmm seems like I didn't need that patch with a fresh build

SGaist

So a completely clean build ?

ltmdum

@SGaist Yeah. I still had to make other changes to get things to work but they were specific to us. I think it must have been the order in which I made the changes that made this seem necessary when they weren't