| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
speed matters. (#64)
This commit adds the HBXX64 and HBXX128 classes for use as a drop-in
replacement for the slow LLMD5 hashing class, where speed matters and
backward compatibility (with standard hashing algorithms) and/or
cryptographic hashing qualities are not required.
It also replaces LLMD5 with HBXX* in a few existing hot (well, ok, just
"warm" for some) paths meeting the above requirements, while paving the way for
future use cases, such as in the DRTVWR-559 and sibling branches where the slow
LLMD5 is used (e.g. to hash materials and vertex buffer cache entries), and
could be use such a (way) faster algorithm with very significant benefits and
no negative impact.
Here is the comment I added in indra/llcommon/hbxx.h:
// HBXXH* classes are to be used where speed matters and cryptographic quality
// is not required (no "one-way" guarantee, though they are likely not worst in
// this respect than MD5 which got busted and is now considered too weak). The
// xxHash code they are built upon is vectorized and about 50 times faster than
// MD5. A 64 bits hash class is also provided for when 128 bits of entropy are
// not needed. The hashes collision rate is similar to MD5's.
// See https://github.com/Cyan4973/xxHash#readme for details.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The unsigned index arithmetic was problematic in that case.
|
|
|
|
Since LLSDSerialize::SIZE_UNLIMITED is negative, passing that through unsigned
size_t parameters could result in peculiar behavior.
|
|
and use it to replace dubious loops in asLLSD() and trimEmpty().
|
|
|
|
|
|
Introduce LLSD template constructors and assignment operators to disambiguate
construction or assignment from any integer type to Integer, likewise any
floating point type to Real. Use new narrow() function to validate
conversions.
For LLSD method parameters converted from LLSD::Integer to size_t, where the
method previously checked for a negative argument, make it now check for
size_t converted from negative: in other words, more than S32_MAX. The risk of
having a parameter forced from negative to unsigned exceeds the risk of a
valid length or index over that max.
In lltracerecording.cpp's PeriodicRecording, now that mCurPeriod and
mNumRecordedPeriods are size_t instead of S32, defend against subtracting 1
from 0.
Use narrow() to validate newly-introduced narrowing conversions.
Make llclamp() return the type of the raw input value, even if the types of
the boundary values differ.
std::ostream::tellp() no longer returns a value we can directly report as a
number. Cast to U64.
|
|
|
|
size_t conversions
|
|
conversions
|
|
llssize is for a function parameter that should accept a size or index
(derived from size_t, which is 64 bits in a 64-bit viewer) but might need to
go negative for flag values. We've historically used S32 for that purpose, but
Xcode 14.1 complains about trying to pass size_t to S32.
narrow() is a template function that casts a wider type (e.g. size_t or
llssize) to a narrower type (e.g. S32 or U32), with validation in
RelWithDebInfo builds. It verifies (using assert()) that the value being
truncated can in fact fit into the target type.
|
|
|
|
It's a little distressing how often we have historically coded S32 or U32 to
pass a length or index.
There are more such assumptions in other viewer subdirectories, but this is a
start.
|
|
|
|
|
|
|
|
# Conflicts:
# indra/newview/llmodelpreview.h
|
|
|
|
|
|
According to bugsplat get_thread_recorder was null
Replaced apr based LLThreadLocalPointer with thread_local
|
|
# Conflicts:
# doc/contributions.txt
# indra/newview/llviewercontrol.cpp
|
|
|
|
|
|
The work queue callback binds "this". This is deemed safe due to current dependencies, but see the associated comment in the return callback. There was some trial and error to get a this-binded lambda to compile.
Due to LLVorbisDecodeState writing to disk off-thread, limit audio decodes proportional to general worker thread count. Guess the thread count for now.
|
|
(cherry picked from commit 41d6a0e222241606c317281e2f0b211e16813dd5)
|
|
# Conflicts:
# indra/llprimitive/llmodel.cpp
# indra/llprimitive/llmodel.h
# indra/newview/llappviewer.cpp
# indra/newview/llappviewer.h
|
|
SL-15937 - adjust memory limit / cache code
Approved-by: Andrey Lihatskiy
Approved-by: Vir Linden
|
|
# Conflicts:
# autobuild.xml
# indra/cmake/LLCommon.cmake
# indra/llcommon/CMakeLists.txt
# indra/llrender/llgl.cpp
# indra/newview/llappviewer.cpp
# indra/newview/llface.cpp
# indra/newview/llflexibleobject.cpp
# indra/newview/llvovolume.cpp
|
|
|
|
|
|
# Conflicts:
# indra/llcommon/llsdutil.cpp
# indra/newview/VIEWER_VERSION.txt
# indra/newview/lldrawpoolalpha.cpp
# indra/newview/lldrawpoolwater.cpp
|
|
# Conflicts:
# indra/newview/app_settings/settings.xml
# indra/newview/llfloatersearch.cpp
# indra/newview/llgroupactions.cpp
# indra/newview/llvovolume.cpp
|
|
|
|
source of the real crash for when the viewer inevitably crashes later.
|
|
builds.
|
|
# Conflicts:
# autobuild.xml
# doc/contributions.txt
# indra/cmake/GLOD.cmake
# indra/llcommon/tests/llprocess_test.cpp
# indra/newview/VIEWER_VERSION.txt
# indra/newview/lldrawpoolavatar.cpp
# indra/newview/llfloatermodelpreview.cpp
# indra/newview/llmodelpreview.cpp
# indra/newview/llviewertexturelist.cpp
# indra/newview/llvovolume.cpp
# indra/newview/viewer_manifest.py
|
|
# Conflicts:
# indra/llaudio/llstreamingaudio_fmodstudio.cpp
# indra/newview/llviewerregion.cpp
|
|
|
|
# Conflicts:
# autobuild.xml
# indra/llcommon/llsys.cpp
# indra/newview/app_settings/key_bindings.xml
# indra/newview/llfloatereditextdaycycle.cpp
|
|
assertions around ref counting and (hack) fix crash on shutdown from dangling texture reference (reduced to 1 dangling texture from several hundred, can't find the remaining reference).
|
|
SL-14961 works better for windows than rethrow
|
