Граф коммитов

7543 Коммитов

Автор SHA1 Сообщение Дата
dbaron%dbaron.org 90bc11aa89 Root the new string across the call to js_NewObject, etc. b=331793 r=brendan 2006-03-27 23:45:56 +00:00
dbaron%dbaron.org c7844c2cd8 Root the result of js_NewXML across js_GetXMLObject. b=331786 r=brendan 2006-03-27 23:43:40 +00:00
brendan%mozilla.org a69a747947 MIPS little-endian porting fix from Ian Jackson and Thiemo Seufer, via Mike Hommey <mh@glandium.org> (331820, r=me). 2006-03-27 05:54:51 +00:00
brendan%mozilla.org 2350ba344e Home sp (and pc) before allocating call-outs (331719, r=mrbkap). 2006-03-26 03:50:34 +00:00
igor%mir2.org dc37a48f53 Bug 331456: Cache of deflated string bytes is per runtime now. To preserve
API compatibility JS_GetStringBytes(JSString *str) calls newly introduced
js_GetGCStringRuntime(JSString *str) to extract JSRuntime* instance based
on the layout of GC structures. r=brendan
2006-03-25 22:55:00 +00:00
brendan%mozilla.org 12081a43a3 Rooting fixes courtesy dbaron (331678, r=mrbkap). 2006-03-25 21:51:57 +00:00
igor%mir2.org 373a0e0bf8 Bug 331598: Cleanup/simplifications after landing of patch for recursion-free GC from bug 324278. 2006-03-25 20:21:31 +00:00
brendan%mozilla.org c0c2ef84b1 Fix silly null ptr deref (331664, r=mrbkap). 2006-03-25 19:24:44 +00:00
mrbkap%gmail.com bd2900305a Fix a silly typo decompiling the setting of the default xml namespace. bug 331558, patch by Erik Fabert <jerfa@email.it>, r=mrbkap 2006-03-24 18:32:11 +00:00
dbaron%dbaron.org d8b42e96fc Fix C++ comments that I introduced. 2006-03-24 08:32:09 +00:00
dbaron%dbaron.org ace148d610 Workaround brokenness of __builtin_frame_address(0) on gcc 4.1 (as shipped with FC5, at least). b=331436 r=brendan 2006-03-23 23:21:27 +00:00
brendan%mozilla.org b22355da41 Missing OBJECT_TO_JSVAL to fix warning introduced by last checkin. 2006-03-23 21:02:43 +00:00
igor%mir2.org dadb71b144 Bug 330692: GC_MARK_DEBUG-only arguments are removed when GC_MARK_DEBUG is not defined. r=brendan 2006-03-23 11:50:30 +00:00
mrbkap%gmail.com 7a4858081c Don't add unnecessary quotes to the rhs of the descendant operator. bug 321547, r=brendan 2006-03-23 05:36:20 +00:00
mrbkap%gmail.com cb3b07a687 Continue to create an exception even when we're unable to convert a function passed as an argument to a string. This fixes some cases where an exception would become uncatchable. bug 328443, r=brendan 2006-03-23 05:16:39 +00:00
mrbkap%gmail.com c29adb7907 Root the new function across XDR operations to make sure it doesn't get garbage collected. bug 327708, r=brendan 2006-03-23 05:12:27 +00:00
mrbkap%gmail.com 0f68c9e245 Protect against scripts setting __proto__ to null. bug 317250, hopeful-r=brendan 2006-03-23 02:24:45 +00:00
mrbkap%gmail.com 7685d1b7e0 Create the PropertyIterator object with a default parent so that when we attempt to find the PropertyIterator constructor, we look up the default scope chain and not one that we might not have access to. bug 328012, r=brendan 2006-03-23 01:25:21 +00:00
mrbkap%gmail.com f9460f5c90 Always innerize the variables object to ensure that random properties don't get stuck on the outer object. bug 317250, r=brendan 2006-03-22 21:44:31 +00:00
igor%mir2.org bd61ffbe63 Bug 324278: Implementation of GC marking algorithm that uses constant space for
any kind of GC things. r=brendan

The main idea is to put a GC thing to a special "unscanned bag" instead of recursively calling GC_MARK on thing's children when C stack is slow. Then later the code loops through the bag marking the children until the bag is empty.

The unscanned bag is implemented as a linked list of GC arenas where things that belongs to the bug marked with GCF_MARK|GCF_FINAL combination. To avoid long scanning of arenas on the list, the code uses a bitmask per arena to indicate which pages within the arena contains unscanned things and an extra bitmask per page to indicate offset range withing the page of the unscanned things.
2006-03-22 15:38:43 +00:00
mrbkap%gmail.com 309575af1a I didn't mean to check this in. 2006-03-22 07:29:49 +00:00
mrbkap%gmail.com 45b704978a Used standard IO functions in snarf so that things work equally well on Linux and Windows. No bug, patch by Brendan, r=mrbkap 2006-03-22 07:26:31 +00:00
mrbkap%gmail.com dc0e278951 Fixing some silliness from my last checkin, tokenstr takes a tokentype, not the token itself. r=brendan 2006-03-22 06:23:59 +00:00
mrbkap%gmail.com 1d4b9a7b0a Handle assignment inside the then part of ternary expressions correctly according to the ECMAScript grammar. bug 330975, r=brendan 2006-03-22 03:00:03 +00:00
brendan%mozilla.org 1d810ceadf More Windows warning fixes. 2006-03-22 01:39:40 +00:00
brendan%mozilla.org c37c9150ee Warning fix for MSVC. 2006-03-22 01:37:27 +00:00
cls%seawood.org 0828b79df3 Fix mingw cross-compile bustage.
Bug #322578 r=mark sr=bsmedberg
2006-03-21 03:04:54 +00:00
mrbkap%gmail.com 5c696de243 Don't forget to throw syntax errors for missing parentheses or brackets. bug 331127, r=brendan 2006-03-20 20:57:51 +00:00
mrbkap%gmail.com f60971fc05 Don't pollute String.prototype. bug 330976, r=brendan 2006-03-20 20:15:15 +00:00
bclary%bclary.com 410c7162e5 Regression test for bug 329530, by Erik Fabert 2006-03-20 19:55:22 +00:00
bclary%bclary.com 0e5f054b05 Regression test for bug 330976, by Blake Kaplan 2006-03-20 19:05:49 +00:00
bclary%bclary.com 4b20c14c15 Regression test for bug 330951, by Pawel Chmielowski 2006-03-20 00:40:45 +00:00
dbaron%dbaron.org 4a829838f0 Use __builtin_frame_address instead of setjmp to get the frame pointer. b=323853 r=brendan 2006-03-19 01:27:17 +00:00
igor%mir2.org cdc28d9089 Bug 330951: Patch from Pawel Chmielowski to fix a regression that I introduced in bug 311515. 2006-03-18 23:36:04 +00:00
igor%mir2.org a254e22506 Bug 330951: Asserting that string arguments str1 and str2 arguments are not NULLs in js_CompareStrings and js_EqualStrings.
Previously passing str1 and str2 set to NULL did not crash as str1 and str2 compared equals as pointers and the functions returned true without accessing *str1 or *str2. In turn it allowed for the regression from bug 311515 causing this bug to survive much longer then it should. r=brendan.
2006-03-18 23:29:15 +00:00
igor%mir2.org 14a2473a5d Bug 330812: In array_sort clear only those elements of temporary vector that
would be included in the rooted set. It provides better performance for huge
mostly spare arrays. r=mrbkap
2006-03-17 22:09:46 +00:00
nboyd%atg.com e61b68c185 Add another user. 2006-03-17 16:16:29 +00:00
igor%mir2.org 5f57130e34 Bug 311515: Fixing misspellings in comments from the previous commit. 2006-03-16 08:04:28 +00:00
igor%mir2.org cba34b6b55 Bug 311515: array_sort sorts undefined value and holes via counting them as according to the Ecma standard holes are bigger then undefined values which are bigger then anything else. This allows to simplify sort_compare and shrink jsarray.c code while making sorting of spare arrays faster. r=mrbkap 2006-03-15 20:49:09 +00:00
igor%mir2.org 6f1486e3be Bug 328664: In js_DecompileValueGenerator search the stack for the value to decompile from the top to bottom to pick up the copy and corresponding code that was pushed to the stack most recently. r=brendan 2006-03-15 16:44:24 +00:00
daumling%adobe.com d5cca305bd Bug 330169: Replaced early exits in ParseNodeToXML() with a jump to skip_child so js_LeaveLocalRootScope() can be called correctly. r=mrbkap 2006-03-15 09:23:36 +00:00
bryner%brianryner.com 4cd1e2b280 Remove dependency on nsIClassInfo.h from nsISupports.h (bug 330420). This adds a new nsIClassInfoImpl.h file which can be included to get the CI implementation macros. Also, removes unneeded inclusion of nsIProgrammingLanguage.h from nsIClassInfo.h. r=darin. 2006-03-15 04:59:42 +00:00
timeless%mozdev.org 25c4968b91 Bug 329530 Out of memory crash when calling fn.toString where fn is a deeply nested function
patch by jerfa@email.it r=brendan
2006-03-13 17:00:18 +00:00
bclary%bclary.com 64a948ecc8 Regression test for bug 301574, fix by shutdown 2006-03-13 02:45:26 +00:00
timeless%mozdev.org 718a68e83c Bug 330098 XPCCallContext::~XPCCallContext is still wiping out newborn roots causing crashes under [@ js_FinalizeObject] because AllocSlots is calling gc and causing its caller (js_NewObject)'s obj to be destroyed before it's stable
r=mrbkap sr=bz
2006-03-13 00:24:06 +00:00
timeless%mozdev.org a74aa18941 Bug 326961 JS_BeginRequest should JS_ASSERT(cx->thread == js_GetCurrentThreadId())
r=brendan r=mrbkap
2006-03-12 09:38:06 +00:00
timeless%mozdev.org 376cbca96e Bug 329749 Did JS_SealObject mean to call JS_UNLOCK_OBJ instead of JS_UNLOCK_SCOPE ?
r=brendan
2006-03-12 09:35:52 +00:00
mrbkap%gmail.com 390716870f VC8 doesn't like the copyright sign in jscompat.h. Issue found by jshin, bug 328589, r=gerv/mrbkap 2006-03-11 04:05:13 +00:00
mrbkap%gmail.com 68ad7cb404 Use __declspec(dllexport/dllimport) on OS/2. Patch from Andy Willis <abwillis1@gmail.com>, bug 301064, r=mrbkap 2006-03-11 01:35:21 +00:00
timeless%mozdev.org e0b0545c82 Bug 329754 three coverity complaints about jsj_JSObject.c
r=mrbkap sr=bz
2006-03-10 20:44:48 +00:00