The "static-analysis-bsmedberg" tinderbox went red this morning for two cycles in a row (so far), with the last line in both logs being: { /builds/gcc-dehydra/installed/bin/g++ -o nsHtml5NamedCharacters.o -c -I../../dist/system_wrappers -include /builds/static-analysis-buildbot/slave/full/build/config/gcc_hidden.h -DMOZILLA_INTERNAL_API -DOSTYPE=\"Linux2.6.27.5-41.fc9\" -DOSARCH=Linux -I/builds/static-analysis-buildbot/slave/full/build/parser/html -I. -I../../dist/include -I../../dist/include/nsprpub -I/builds/static-analysis-buildbot/slave/full/build/obj-buildbot/dist/include/nspr -I/builds/static-analysis-buildbot/slave/full/build/obj-buildbot/dist/include/nss -I/builds/static-analysis-buildbot/slave/full/build/parser/html/../../content/base/src -fPIC -fno-rtti -fno-exceptions -Wall -Wpointer-arith -Woverloaded-virtual -Wsynth -Wno-ctor-dtor-privacy -Wno-non-virtual-dtor -Wcast-align -Wno-invalid-offsetof -Wno-long-long -pedantic -fno-strict-aliasing -fshort-wchar -pthread -pipe -DDEBUG -D_DEBUG -DDEBUG_bsmedberg -DTRACING -g -fno-inline -fplugin=/builds/gcc-dehydra/dehydra/gcc_treehydra.so -fplugin-arg='/builds/static-analysis-buildbot/slave/full/build/config/static-checking.js --topsrcdir=/builds/static-analysis-buildbot/slave/full/build --objdir=../.. --dehydra-modules=/builds/static-analysis-buildbot/slave/full/build/xpcom/analysis/final.js,/builds/static-analysis-buildbot/slave/full/build/layout/generic/frame-verify.js --treehydra-modules=/builds/static-analysis-buildbot/slave/full/build/xpcom/analysis/outparams.js,/builds/static-analysis-buildbot/slave/full/build/xpcom/analysis/stack.js,/builds/static-analysis-buildbot/slave/full/build/xpcom/analysis/flow.js,/builds/static-analysis-buildbot/slave/full/build/js/src/jsstack.js ' -Os -freorder-blocks -fno-reorder-functions -DMOZILLA_CLIENT -include ../../mozilla-config.h -Wp,-MD,.deps/nsHtml5NamedCharacters.pp /builds/static-analysis-buildbot/slave/full/build/parser/html/nsHtml5NamedCharacters.cpp } There's no obvious mention of any errors, so this is probably just timing out or something. (bsmedberg says in IRC "it's likely that the analysis is actually ilooping or doing some explosive algorithm in the new HTML5 parser code") Logs: http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1246899162.1246903745.28268.gz static-analysis-bsmedberg on 2009/07/06 09:52:42 http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1246908297.1246912864.5381.gz static-analysis-bsmedberg on 2009/07/06 12:24:57 Blame range is: http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=b4fee3813956&tochange=33475a1b5746 In that range, bug 502103 is the only html5-related change, so it's probably the change that triggered the timeout.

Strange. The patch from bug 502103 had nothing to do with nsHtml5NamedCharacters.cpp. However, nsHtml5NamedCharacters.cpp caused excessive build times with MSVC until compiler optimizations were turned off for initializeStatics (see bug 501082). The real fix for making nsHtml5NamedCharacters.cpp better is bug 502176. Can static analysis be made analyze nsHtml5NamedCharacters::initializeStatics less for now?

I did some local debugging: it seems that the static analysis scripts weren't directly involved in the slowdown compiling nsHtml5NamedCharacters::initializeStatics. The slowdown happens before the input_end callback, but the build does eventually complete. I've bumped up the timeout and am going to cross my fingers that this is fixed or a fluke.