You are not logged in.
Pages: 1
I downgraded from 6.0 to 5.44 but have the same result.
No matter which screensaver I choose, the display says 'No preview available' and the journal has this message:
Jun 01 17:53:32 asgl systemd[1]: systemd-coredump@11-17350-0.service: Deactivated successfully.
Jun 01 17:53:32 asgl systemd-coredump[17351]: [LNK] Process 17339 (gflux) of user 1000 dumped core.
Stack trace of thread 17339:
#0 0x00007fd8879dcd22 raise (libc.so.6 + 0x3cd22)
#1 0x00007fd8879c6862 abort (libc.so.6 + 0x26862)
#2 0x000055b97a19e608 n/a (gflux + 0xb608)
#3 0x000055b97a19c01b n/a (gflux + 0x901b)
#4 0x000055b97a19d94e n/a (gflux + 0xa94e)
#5 0x000055b97a199983 n/a (gflux + 0x6983)
#6 0x00007fd8879c7b25 __libc_start_main (libc.so.6 + 0x27b25)
#7 0x000055b97a19a37e n/a (gflux + 0x737e)
This is a recent installation of Arch so I can't say that xscreensaver has ever worked on this machine.
Using NVIDIA 390.143 for GeFiorce GTX 560M
Offline
Can you run /usr/lib/xscreensaver/xmatrix (non-GL XSS hack)?
If not, try to "strace /usr/lib/xscreensaver/xmatrix" - the backtrace has no debug symbols what is going to make it hard to determine the cause from it alone.
But it aborts, so this is a "controlled" error.
Online
Can you run /usr/lib/xscreensaver/xmatrix (non-GL XSS hack)?
Yes. All the X* screensavers play.
Also: Decay Screen, Cynosure, Crystal... There may be others. Most do not play.
If not, try to "strace /usr/lib/xscreensaver/xmatrix" - the backtrace has no debug symbols what is going to make it hard to determine the cause from it alone.
But it aborts, so this is a "controlled" error.
Yes . At timeout the screensaver puts a big red font saying xmatrix sigabort
"strace /usr/lib/xscreensaver/glslideshow" produces many cryptic lines and ends with a SIGABRT message
Last edited by holiday (2021-06-02 17:07:08)
Offline
Please post the output of "glxinfo" and also of
strace /usr/lib/xscreensaver/glslideshow 2>&1 | curl -F 'f:1=<-' ix.io
Online
Offline
write(2, "glslideshow: eglCreatePlatformWi"..., 81glslideshow: eglCreatePlatformWindowSurface failed: window 0x4e00002 visual 0x24
You even got a nice error message, telling you what's broken.
So
Please post the output of "glxinfo"
You can also add your xorg log, because I'm pretty sure the GL installation is broken.
Online
Please post the output of "glxinfo"
You can also add your xorg log, because I'm pretty sure the GL installation is broken.
Offline
Nope. GLX is fine.
And: it's actually reproducible for me.
https://www.jwz.org/xscreensaver/changelog.html
X11: Use EGL instead of GLX when available.
I assume it's somehow incompatible w/ the nvidia implementation and there's no obvious way to enforce glx w/o rebuilding xscreensaver and even then, whole files are wrapped in an "#if HAVE_EGL" or so preproc condition.
Edit: it's not a false positive, skipping the abort() will cause a later segfault.
eglgears and egltri create a 0x27 visual, the screensaver hacks choose 0x24 - however forcing 0x27 in visual-gl.c doesn't change anything but the visual id.
Edit #2: it's the specific API
d->egl_surface = eglCreateWindowSurface (d->egl_display,
d->egl_config,
window, NULL);
works.
=> I've mailed jwz, hope for an upstream patch.
Last edited by seth (2021-06-03 16:19:27)
Online
XSS patch:
--- xlock-gl-utils.c 2021-06-03 18:14:41.428137527 +0200
+++ xlock-gl-utils.c.orig 2021-02-16 17:32:29.000000000 +0100
@@ -130,9 +130,9 @@ init_GL(ModeInfo * mi)
abort();
}
- d->egl_surface = eglCreateWindowSurface (d->egl_display,
+ d->egl_surface = eglCreatePlatformWindowSurface (d->egl_display,
d->egl_config,
- window, NULL);
+ &window, NULL);
if (! d->egl_surface)
{
fprintf (stderr, "%s: eglCreatePlatformWindowSurface failed:"
Edit: the 465xx drivers are reported to be not affected.
Last edited by seth (2021-06-06 21:12:39)
Online
Pages: 1