JSTests/stress/joined-strings-should-not-exceed-max-string-length.js introduced in r207849 and JSTests/stress/string-prototype-replace-should-throw-out-of-memory-error-when-using-too-much-memory.js introduced in r207861. Both of these test consume huge memory, so they should be skipped on memory limited devices. test1 ------ //@ runFTLNoCJIT //@ largeHeap test2 ------ //@ runFTLNoCJIT //@ slow! //@largeHeap The original intention was to skip these tests on memory limited devices, but unfortunately run-jsc-stress-tests script can't handle these annotations properly. RJST doesn't have any complex logic, it simply executes everything after "//@" magic annotation. But runFTLNoCJIT, runDefault, etc. doesn't know if you put largeHeap annotation too to the same test file. There were a similar fix not so long ago in bug162386 . I'm going to land the same workaround to this bug to unbreak JSCOnly ARM bots. But please let me know if you know a better and general fix for this RJST bug/feature.
Created attachment 293150 [details] Patch for landing
(In reply to comment #0) > JSTests/stress/joined-strings-should-not-exceed-max-string-length.js > introduced in r207849 and > JSTests/stress/string-prototype-replace-should-throw-out-of-memory-error- > when-using-too-much-memory.js > introduced in r207861. Both of these test consume huge memory, so they > should be skipped on memory > limited devices. note: - JSTests/stress/joined-strings-should-not-exceed-max-string-length.js consumes 1.1Gb memory on X86_64 - JSTests/stress/string-prototype-replace-should-throw-out-of-memory-error- > when-using-too-much-memory.js consumes 6.5Gb memory on X86_64
Comment on attachment 293150 [details] Patch for landing Clearing flags on attachment: 293150 Committed r208047: <http://trac.webkit.org/changeset/208047>
All reviewed patches have been landed. Closing bug.