Hello
Here's a better regression fix for the dinput-scancode patchset (the 0001 patch of that patchset caused NFS Underground to segfault since Wine 9.15 with the addition of a keycode loop in keyboard_create_device() because of DIPROP_SCANCODE stuff); I modified the existing 0001 patch to solve this by copying the duplicate object checks to that new loop (maybe I should create a separate 0003 patch to show the changes better?)
Hopefully the patch can be merged before 9.16 to not re-introduce that regression after it has been worked around in wine-staging 9.15
Thanks
On Thursday, 22 August 2024 14:23:13 CDT DodoGTA GT wrote:
Hello
Here's a better regression fix for the dinput-scancode patchset (the 0001 patch of that patchset caused NFS Underground to segfault since Wine 9.15 with the addition of a keycode loop in keyboard_create_device() because of DIPROP_SCANCODE stuff); I modified the existing 0001 patch to solve this by copying the duplicate object checks to that new loop (maybe I should create a separate 0003 patch to show the changes better?)
Rémi, can you comment on this? Since you're the author of the original patch, I'd appreciate your review before adding this one into Wine-Staging.
And, if correct, does patch 0002 need a similar fix?
--Zeb
On 8/22/24 23:17, Elizabeth Figura wrote:
On Thursday, 22 August 2024 14:23:13 CDT DodoGTA GT wrote:
Hello
Here's a better regression fix for the dinput-scancode patchset (the 0001 patch of that patchset caused NFS Underground to segfault since Wine 9.15 with the addition of a keycode loop in keyboard_create_device() because of DIPROP_SCANCODE stuff); I modified the existing 0001 patch to solve this by copying the duplicate object checks to that new loop (maybe I should create a separate 0003 patch to show the changes better?)
Rémi, can you comment on this? Since you're the author of the original patch, I'd appreciate your review before adding this one into Wine-Staging.
And, if correct, does patch 0002 need a similar fix?
--Zeb
Yes, the Aidas patch looks correct, and yes probably 0002 will need the same kind of thing. Both loops should be pretty much identical. I'm attaching a change that factors them together first, with the other patches applied on top.