-
Notifications
You must be signed in to change notification settings - Fork 667
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
libffi on SPARC V9 have problem calling functions with floating point number arguments #778
Comments
The resolution for this particual issue seems to be add
for Line 385 in ac598b7
But there seems to be some other SPARC problems too... (gnome-shell has still some issues even it's now able to start). |
I see that On the other hand |
Thanks for reporting this, I think I've seen this as well but I wasn't sure enough yet if it was a libffi bug (https://bugs.gentoo.org/882071) but I think it must be the same. |
cc @rth7680 (commit author), although not sure if you're involved with libffi nowadays. |
I have looked at this again and now with my initial finding in #778 (comment) I don't see any other obvious problems (I believe originally there was problem with Gnome unlocking for example). So, maybe following patch is enough?
@thesamesam can you please test whether it solves your problems? |
Setting FP_ARGS for FFI_TYPE_UINT32 is obviously wrong. |
Ok, it sounds right. I have updated the patch. But with my test case of Gnome (gjs) it doesn't seem to make any difference...
|
Thanks, I'll try the second patch first. Wrt no difference - you mean both patches fix your issue, or you mean the second patch doesn't fix your gjs problem? (Hopefully the former..) |
Yes, the seond patch works for me too. I wasn't able to run gjs tests (which is probably your test case) as there is some unrelated Python problem (I think). |
gjs tests before, with the failing GIMarshalling test:
After, all tests pass:
Thanks a bunch! Would you mind making a PR for it? |
This fixes the gjs test suite on sparc. Tests continue to pass on libffi and gjs tests now completely pass (previously GIMarshalling failed). The fix is obvious so I don't see much of a need to wait until it's merged upstream, and it only affects sparc. Bug: libffi/libffi#778 Closes: https://bugs.gentoo.org/882071 Signed-off-by: Sam James <sam@gentoo.org>
Fixed by #802. |
I see the problem on Solaris SPARC V9 with
/usr/bin/gjs
when it callsclutter_actor_set_size (ClutterActor *self, gfloat width, gfloat height);
https://github.com/GNOME/mutter/blob/d643eb5c6fe50e7f1afffda0e8747a87f668a799/clutter/clutter/clutter-actor.c#L9473
Note the floating point number arguments. When I add some debug printfs there I see that it gets just zeros.
This is regression. Old version 3.2.1 worked well. The first bad commit was: 2b27890
(but I know it was major rewrite up and there were other follow up SPARC releated commits)
The text was updated successfully, but these errors were encountered: