Arch Linux User Repository

I finally had time to investigate the issue. It looks like that as of 11.5, the DesktopVideoHelper.service has to be running for the capture card to work properly.

Patch for kernel 5.15, which deprecated alloc_tty_driver and put_tty_driver:

--- package.orig/Blackmagic_Desktop_Video_Linux_12.2/other/x86_64/desktopvideo-12.2a12-x86_64/usr/src/blackmagic-12.2a12/blackmagic_serial.c    2021-11-04 17:18:37.053086911 +0100
+++ package.new/Blackmagic_Desktop_Video_Linux_12.2/other/x86_64/desktopvideo-12.2a12-x86_64/usr/src/blackmagic-12.2a12/blackmagic_serial.c 2021-11-04 17:30:49.791835250 +0100
@@ -613,7 +613,7 @@
    int ret;
    struct tty_driver *driver;

-   driver = alloc_tty_driver(BLACKMAGIC_SERIAL_MINORS);
+   driver = tty_alloc_driver(BLACKMAGIC_SERIAL_MINORS, 0);
    if (!driver)
        return -ENOMEM;

@@ -639,7 +639,7 @@

 abort:
    printk(KERN_ERR "failed to register blackmagic serial driver");
-   put_tty_driver(driver);
+   tty_driver_kref_put(driver);
    return ret;
 }

@@ -648,6 +648,6 @@
    if (blackmagic_tty_driver)
    {
        tty_unregister_driver(blackmagic_tty_driver);
-       put_tty_driver(blackmagic_tty_driver);
+       tty_driver_kref_put(blackmagic_tty_driver);
    }
 }
diff --unified --recursive --text --color package.orig/Blackmagic_Desktop_Video_Linux_12.2/other/x86_64/desktopvideo-12.2a12-x86_64/usr/src/blackmagic-io-12.2a12/bmio_serial.c package.new/Blackmagic_Desktop_Video_Linux_12.2/other/x86_64/desktopvideo-12.2a12-x86_64/usr/src/blackmagic-io-12.2a12/bmio_serial.c
--- package.orig/Blackmagic_Desktop_Video_Linux_12.2/other/x86_64/desktopvideo-12.2a12-x86_64/usr/src/blackmagic-io-12.2a12/bmio_serial.c   2021-11-04 17:18:37.053086911 +0100
+++ package.new/Blackmagic_Desktop_Video_Linux_12.2/other/x86_64/desktopvideo-12.2a12-x86_64/usr/src/blackmagic-io-12.2a12/bmio_serial.c    2021-11-04 17:29:38.749443820 +0100
@@ -304,7 +304,7 @@
 {
    int ret;

-   struct tty_driver* serial = alloc_tty_driver(MAX_SERIAL_LINES);
+   struct tty_driver* serial = tty_alloc_driver(MAX_SERIAL_LINES, 0);
    if (!serial)
        return -ENOMEM;

@@ -324,7 +324,7 @@
    ret = tty_register_driver(serial);
    if (ret)
    {
-       put_tty_driver(serial);
+       tty_driver_kref_put(serial);
        return ret;
    }

@@ -340,7 +340,7 @@
    if (serial_driver)
    {
        tty_unregister_driver(serial_driver);
-       put_tty_driver(serial_driver);
+       tty_driver_kref_put(serial_driver);
        serial_driver = NULL;
    }
 }

awesome, also working here on 5.14.3-arch1-1 -- happy to be able to update the system, was holding off until now since I rely on the decklink drivers, using it everyday -- thanks for the package update! FWIW it did require a reboot for the device to be found by BlackmagicDesktopVideoSetup.

Can confirm it works for me too with 5.14.2.arch1-2. Thanks for updating the PKGBUILD.

EDIT: Though, the audio stutters quite badly. I'll have to see if I can get that fixed...

EDIT2: happens on Windows too, I think it may have to do with PCI bandwidth available. I get a warning about dropped frames when I record with MediaExpress. But there's not as many dropped frames when I record a 720p60 signal vs 1080p60.

EDIT3: yeah that was it. Unplugging the TV tuner card I never use gave me the PCI bandwidth the card needs to work properly.

edit - sorry, I don't know the half markdown syntax here. deleted my PKGBUILD patch.

edit2: sorry for the lame workaround - uploaded a PKGBUILD gist here: https://gist.github.com/frostworx/6f1c12bdf5c0f319bcbd0ef18e3e301b

edit3: I can confirm that the decklink works fine with 5.14.2-xanmod1-1 with the patched drivers

thx for the quick reply! just gave it a quick try. The patch applies fine, but when manually building the build fails with "bmio_serial.c:174:28: Fehler: Initialisierung von »unsigned int ()(struct tty_struct )« von inkompatiblem Zeigertyp »int ()(struct tty_struct )« [-Werror=incompatible-pointer-types]" (the other module blackmagic-12.1a9 has the same error). the Werror can be disabled by adding "-Wno-error=incompatible-pointer-types" to the EXTRA_CFLAGS in the Makefiles. Both modules build fine then. Haven't tested functionality, as I'm almost off for a short trip \o/ :) Good luck and thanks a lot for maintaining the PKGBUILD! :)

thx, I'll try the patch later today after work. If it builds on my machine I'll push a new revision to the decklink package.

fyi: a blackmagic dev posted a patch in the thread mentioned before here: https://forum.blackmagicdesign.com/viewtopic.php?f=12&t=144876#p788448 haven't tested it yet

fyi, upstream forum issue thread: https://forum.blackmagicdesign.com/viewtopic.php?f=12&t=144876

About a month ago I had sent in my card to their technical support team, thinking it might have been a hardware issue. The other day I finally heard back from them, and they said it was working on Windows. They said they did not have a Linux machine to test with (despite advertising Linux support on the box) and were not able (allowed?) to run a live DVD. They are now shipping it back to me.

@scaramanga I just coincidentally found out that the drivers are currently broken (I barely use the card) and therefore found your comment. Thanks for the heads-up & contacting upstream! :)