2025.03 Cherry pick from Maint B #3708
Conversation
# Conflicts: # indra/newview/lloutfitslist.cpp
# Conflicts: # indra/newview/llfloateremojipicker.cpp # indra/newview/skins/default/xui/en/floater_fast_timers.xml
# Conflicts: # indra/newview/llviewermedia.cpp
# Conflicts: # indra/newview/llphysicsshapebuilderutil.cpp
Second Life's SOCKS5 proxy has been broken on windows for at least six years due to a conflation of milliseconds and microseconds in the APR timeout value used when attempting to ping the proxy. # Conflicts: # indra/llmessage/llproxy.cpp
# Conflicts: # indra/newview/llfilepicker.cpp
Ensure folder creation callbacks remain alive # Conflicts: # indra/newview/llfloatercreatelandmark.cpp # indra/newview/llfloatercreatelandmark.h
# Conflicts: # indra/newview/llfloaterpreferencesgraphicsadvanced.cpp
Marketplace deliberately doesn't allow creating folders and the "New folder from selected" is a way of creating a new folder then moving selected items inside, therefore shouldn't be allowed.
Found by running with -fsanitze=thread Suggestion to avoid accessing invalid memory: In both cases memory will be allocated by can be accessed beyond bounds. In LLPolyMesh it can be off by at least one (+x%2). Though I am not even sure if even in best case it always will be a multiple of 16. In LLViewerJointMesh::updateFaceData the code tries to account for padding by, but the allocation in LLPolyMeshSharedData::allocateVertexData is done without any padding. Thus the sizes must not match. Replacing the calls with memcpy as a quick fix to see if the error goes away fixed address sanitzer complaining. It is up to debate if memcpy is a good replacement. LLVector4a::memcpyNonAliased16 was invented for performance. But on the other hand one could argue that nowadays every stdlib maintainer will very heavily optmize functions like memcpy themselves and could take advantage of CPU features the old LL implementation does not take into account. AVX comes to mind. In any case did I not measure any of this.
# Conflicts: # indra/newview/llviewermedia.cpp
delete once per session, not once per instance # Conflicts: # indra/newview/llviewermedia.cpp
# Conflicts: # indra/llxml/llxmltree.cpp
attachment list gets filled with object ids, not attachment ids
The viewer now prefers to load inventory via the capability rather than over the deprecated Xfer system, though both are still supported. # Conflicts: # indra/newview/llviewerobject.h
|
ea7b63f - this one will likely need some love post-merge. We have a new reflection probe slider in AGP. |
|
Other than the two commits above - that's about all I really see in here to be careful about. The physics display bit is most likely something we took elsewhere and is already, in some form, in shipping code. But I'd like some more clarity around that one. For the AGP - I'm gonna ask Steeltoe if he can come up with some alternative designs. That floater exploded in terms of settings due to PBR, Featurettes, and more recent features. If I can get just a bit more clarity around the physics display bit, then I'll give this a thumbs up. |
Looks like it was already fixed in the base branch (hence the conflict), and git just picked up the rest after resolving it. |
|
@marchcat That's good to know. That resolves the biggest concern I had here. Other than that - just need to figure out what to do with the AGP and I think we're good here. |
A small collection of fixes cherry-picked from Maint B.
This PR excludes any changes related to Linux; they will be added separately.