Created attachment 367724 [details, diff] spidermonkey-24.2.0-stack-grows-up.patch Patch fixes these errors: /dev/shm/portage/dev-lang/spidermonkey-24.2.0/work/mozjs-24.2.0/js/src/gc/RootMarking.cpp:323:20: error: invalid conversion from 'uintptr_t {aka unsigned int}' to 'uintptr_t* {aka unsigned int*}' [-fpermissive] /dev/shm/portage/dev-lang/spidermonkey-24.2.0/work/mozjs-24.2.0/js/src/jsapi.cpp:868:5: error: 'nativeStackLimit' was not declared in this scope
Tests fail with this patch though, yes? Same/similar failures as to what you see with sm17 (bug 502124) ?
Two tests fail, but I was fixing a compile error here. From a 24.2.0 build log: TEST-UNEXPECTED-FAIL | testOOM | Failed to initialize. TEST-UNEXPECTED-FAIL | testGCOutOfMemory | Failed to initialize. [...] 2 unexpected failures.
Thanks! I know about the purpose of the patch, i just wanted to check if the failing tests relate to the stack-growth direction in any way, to see if the patch should be expanded to fix them somehow, too.
testOOM fails predictably(!) in: JS_SetGCParameter(rt, JSGC_MAX_BYTES, (uint32_t)-1); because you can't reliably do (uint32_t)-1 and just hope the trick works on every platform.
(In reply to Jeroen Roovers from comment #4) > testOOM fails predictably(!) in: > > JS_SetGCParameter(rt, JSGC_MAX_BYTES, (uint32_t)-1); > > because you can't reliably do (uint32_t)-1 and just hope the trick works on > every platform. Nothing to do with stack direction; this is due to endianness.
+ 05 May 2014; Ian Stakenvicius (_AxS_) <axs@gentoo.org> + +files/spidermonkey-24-upward-growing-stack.patch, + spidermonkey-24.2.0-r1.ebuild: + fix compilation on hppa (bug 497900) might as well keep this open until the tests are sorted, though
(In reply to Ian Stakenvicius from comment #6) > + 05 May 2014; Ian Stakenvicius (_AxS_) <axs@gentoo.org> > + +files/spidermonkey-24-upward-growing-stack.patch, > + spidermonkey-24.2.0-r1.ebuild: > + fix compilation on hppa (bug 497900) > > > might as well keep this open until the tests are sorted, though Again, you're confusing a compile failure with an unrelated test failure.
I am, but figured I might as well overload this bug than have a new one for the tests. Of note, though, tests on hake pass clean with the patch committed.