Drake points out that the OS X 64-bit-capable memory-query APIs recommended in
comments by some long-ago maintainer are by now themselves obsolete. He
offered this patch to update us to current macOS memory APIs.
There were two distinct LLMemory methods getCurrentRSS() and
getWorkingSetSize(). It was pointless to have both: on Windows they were
completely redundant; on other platforms getWorkingSetSize() always returned
0. (Amusingly, though the Windows implementations both made exactly the same
GetProcessMemoryInfo() call and used exactly the same logic, the code was
different in the two -- as though the second was implemented without awareness
of the first, even though they were adjacent in the source file.)
One of the actual MAINT-6996 problems was due to the fact that
getWorkingSetSize() returned U32, where getCurrentRSS() returns U64. In other
words, getWorkingSetSize() was both useless *and* wrong. Remove it, and change
its one call to getCurrentRSS() instead.
The other culprit was that in several places, the 64-bit WorkingSetSize
returned by the Windows GetProcessMemoryInfo() call (and by getCurrentRSS())
was explicitly cast to a 32-bit data type. That works only when explicitly or
implicitly (using LLUnits type conversion) scaling the value to kilobytes or
megabytes. When the size in bytes is desired, use 64-bit types instead.
In addition to the symptoms, LLMemory was overdue for a bit of cleanup.
There was a 16K block of memory called reserveMem, the comment on which read:
"reserve 16K for out of memory error handling." Yet *nothing* was ever done
with that block! If it were going to be useful, one would think someone would
at some point explicitly free the block. In fact there was a method
freeReserve(), apparently for just that purpose -- which was never called. As
things stood, reserveMem served only to *prevent* the viewer from ever using
that chunk of memory. Remove reserveMem and the unused freeReserve().
The only function of initClass() and cleanupClass() was to allocate and free
reserveMem. Remove initClass(), cleanupClass() and the LLCommon calls to them.
In a similar vein, there was an LLMemoryInfo::getPhysicalMemoryClamped()
method that returned U32Bytes. Its job was simply to return a size in bytes
that could fit into a U32 data type, returning U32_MAX if the 64-bit value
exceeded 4GB. Eliminate that; change all its calls to getPhysicalMemoryKB()
(which getPhysicalMemoryClamped() used internally anyway). We no longer care
about any platform that cannot handle 64-bit data types.
While we could set major and minor operating system version with the data
returned by GetVersionEx / GetVersion, we set them explicitly as named function
may be altered or removed in the future. In that case, only the build number
would be affected.
Also removed the old way of determing the build number via shell version. This
was only needed for Windows 9x, which aren't supported anymore and can't even
run the viewer.
continued conversion to units system
made units perform type promotion correctly and preserve type in arithmetic
e.g. can now do LLVector3 in units
added typedefs for remaining common unit types, including implicits
cleaning up build
moved most includes of windows.h to llwin32headers.h to disable min/max macros, etc
streamlined Time class and consolidated functionality in BlockTimer class
llfasttimer is no longer included via llstring.h, so had to add it manually in several places
- Adjust whitespaces/include order to make FS source look as close to V/dev as possible.
- Include source in V/dev that's not in FS and comment it out with an annotation.
Observations and broken stuff so far:
- LLSideTray is gone.
- LLBottomTray is gone.
- LLFloaterSnapshot was changed and gave lots of merge conflicts. Flickr upload will be broken.
- LLNearbyChat is derived from LLPanel now. That broke quite some stuff rearding autohiding and docking of chatbar.
- Profile floaters are gone and all web based now.
- Russian translation gave huge mere conflicts with the new russian xml files from LL. Unmergeable. I took the LL files.
- XUI files in general will need some love and testing.
(per Monty code review)
The notion of storing LLMemoryInfo data both as an LLSD::Map and an
LLSD::Array of pair arrays arose from a (possibly misguided) desire to
continue producing stats output into the viewer log in the same order it
always used to be produced. There is no evidence that anyone cares about the
order of those stats in the log; there is no other use case for preserving
order. At Monty's recommendation, eliminate generating and storing the
array-of-pair-arrays form: directly store LLSD::Map.
(per Monty code review)
Explain why we intentionally don't suppress exceptions from boost::regex
objects constructed with string literals. Catch std::runtime_error from
boost::regex_search() and boost::regex_match(); log and return false.
On Mac, where LLMemoryInfo relies on a child process rather than any sort of
internal system API, try to produce more informative LL_WARNS output if
popen() fails to run vm_stat, or if vm_stat terminates with nonzero rc.
For postprocessing these stats, we'll want the time at which they were
captured. We'll want the current framerate too, but handle that at a higher
level.
Ansariel