This was a nice and simple way to get a perf boost in the
interpreter and JM, but JSOP_GETGNAME has the same
information and more. TI doesn't need it, JM technically
doesn't, and IM won't either. We can just do a normal
property lookup during compilation.
Hg commit 804e93f6 includes a zero-length copy of android_npapi.h in
dom/plugins/base/. As far as I can tell this isn't functional; there's a
working copy of of the header in dom/plugins/base/android.
This uses the NS_XPCOM_WILL_SHUTDOWN_OBSERVER_ID notification to produce
the places shutdown messages, emulating what happens in the browser.
--HG--
extra : rebase_source : ddd5bf7973458df14c31f72567b147996b0fac00
otherwise fails with:
ipc/chromium/src/base/platform_file_posix.cc:49: error: 'S_IRUSR' was not declared in this scope
ipc/chromium/src/base/platform_file_posix.cc:49: error: 'S_IWUSR' was not declared in this scope
otherwise fails with:
/usr/include/sys/proc.h:64: error: 'MAXLOGNAME' was not declared in this scope
/usr/include/sys/proc.h:287: error: 'MAXCOMLEN' was not declared in this scope
Don't try to include execinfo.h/use backtrace* funcs on OpenBSD too, otherwise fails with
../../dist/system_wrappers/execinfo.h:3:27: error: execinfo.h: No such file or directory
In constructor 'StackTrace::StackTrace()':
/home/landry/src/mozilla-central/ipc/chromium/src/base/debug_util_posix.cc:125: error: 'backtrace' was not declared in this scope
/home/landry/src/mozilla-central/ipc/chromium/src/base/debug_util_posix.cc: In member function 'void StackTrace::PrintBacktrace()':
/home/landry/src/mozilla-central/ipc/chromium/src/base/debug_util_posix.cc:144: error: 'backtrace_symbols_fd' was not declared in this scope
../../ipc/chromium/src/base/singleton.h:171: error: invalid conversion
from 'base::subtle::AtomicWord*' to 'volatile base::subtle::Atomic32*'
Lifted from www/chromium/patches/patch-base_atomicops_h
Relanding again. I've been unable to understand the regressions and
I can't reproduce them locally anymore. None of the measurements
relavant to this patch change negatively, and we know this is
the direction we want to go. The saner memory behaviour is worth
any small regressions that this might cause.
--HG--
extra : rebase_source : 73f4298fa69c8fc78a60a8c99e551a5554316a93
These just add a bunch of noise to the data because they break some of the assumptions we have.
Currently this doesn't have much impact because decode on draw is disabled.
--HG--
extra : rebase_source : 1022ad9bd43ea5f8be342c05ab857bb44ed7e59c
