As it happens, the change in the LLUUID::combine() algorithm introduced by one
of my previous commits is causing invalid assets creation (seen with
some clothing items, such as Shape and Universal types); obviously, the server
is using the old algorithm for UUID validation purpose of these assets.
This commit reverts LLUUID::combine() code to use LLMD5.
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.
Revert part of "DRTVWR-575: Address review comments on Xcode 14.1 type tweaks."
Crash was reproduced when assigning areastr to llsd, but likely present in other cases of assigning ui strings to llsd (instead of going for lluistring's result directly copy constructor was engaged and either copy or original crashed due to invalid pointers, copy shouldn't have been created).
One could argue that passing a negative index to an LLSD array should do
something other than shrug and reference element [0], but as that's legacy
behavior, it seems all too likely that the viewer sometimes relies on it.
This specific problem arises if the index passed to operator[]() is negative
-- either with the previous Integer parameter or with size_t (which of course
reinterprets the negative index as hugely positive). The non-const
ImplArray::ref() overload checks parameter 'i' and, if it appears negative,
sets internal 'index' to 0.
But in the next stanza, if (index >= existing size()), it calls resize() to
scale the internal array up to one more than the requested index. The trouble
is that it passed resize(i + 1), not the adjusted resize(index + 1).
With a requested index of exactly -1, that would pass resize(0), which would
result in the ensuing array[0] reference being invalid.
With a requested index less than -1, that would pass resize(hugely positive)
-- since, whether operator[]() accepts signed LLSD::Integer or size_t,
resize() accepts std::vector::size_type. Given that the footprint of an LLSD
array element is at least a pointer, the number of bytes required for
resize(hugely positive) is likely to exceed available heap storage.
Passing the adjusted resize(index + 1) should defend against that case.
Henri Beauchamp