80615 2012-03-08 11:13:56 -0800 CopiedSpace::tryAllocateOversize assumes system page size 2012-03-15 16:04:59 -0700 1 1 1 Unclassified WebKit JavaScriptCore 528+ (Nightly build) Unspecified Unspecified RESOLVED FIXED P2 Normal --- 0 litherum webkit-unassigned fpizlo ggaren mhahnenberg webkit.review.bot oldest_to_newest 573975 0 litherum 2012-03-08 11:13:56 -0800 Crash in CopiedSpace::tryAllocateOversize: 1. CopiedSpace::tryAllocateOversize is called with a number of bytes, and a pointer to set (See the end of this email for a stack trace). 2. In order to get the block size, this function calls WTF::roundUpToMultipleOf<s_pageSize>(sizeof(CopiedBlock) + bytes); which attempts to round its argument up to a multiple of s_pageSize. This s_pageSize refers to JSC::CopiedSpace::s_pageSize, declared as a static const in JavaScriptCore/heap/CopiedSpace.h as 4 * KB, or 4096. 3. CopiedSpace::tryAllocateOversize then passes this block size to WTF::PageAllocationAligned::allocate, which immediately checks that this parameter, size, is page-aligned via a call to WTF::isPageAligned(size) 4. WTF::isPageAligned queries the system page size with WTF::pageSize(), which returns WTF::s_pageSize, which has been initialized to the system page size with a call to getpagesize(). On our system, this returns 16384. 5. WTF::isPageAligned returns false because the value that is divisible by 4096 is not divisible by 16384. 6. The assert in WTF::PageAllocationAligned::allocate fails. Eliminating CopiedSpace::s_pageSize in favor of using WTF::s_pageSize would probably be the preferable route. Stack trace: std::size_t WTF::pageSize() bool WTF::isPageAligned(unsigned int) void WTF::PageAllocationAligned::allocate(unsigned int,unsigned int,WTF::OSAllocator::Usage,bool,bool) void JSC::CopiedSpace::tryAllocateOversize(unsigned int,void**) void JSC::CopiedSpace::tryReallocateOversize(void**,unsigned int,unsigned int) void JSC::CopiedSpace::tryReallocate(void**,unsigned int,unsigned int) void JSC::Heap::tryReallocateStorage(void**,unsigned int,unsigned int) bool JSC::JSArray::increaseVectorLength(JSC::JSGlobalData&,unsigned int) void JSC::JSArray::putByIndexBeyondVectorLength(JSC::ExecState*,unsigned int,JSC::JSValue) void JSC::JSArray::putByIndex(JSC::JSCell*,JSC::ExecState*,unsigned int,JSC::JSValue) JSC::EncodedJSValue JSC::arrayProtoFuncPush(JSC::ExecState*) JSC::JSValue JSC::Interpreter::privateExecute(JSC::Interpreter::ExecutionFlag,JSC::RegisterFile*,JSC::ExecState*) 573982 1 litherum 2012-03-08 11:15:44 -0800 I'd like to claim this bug. 575073 2 131073 litherum 2012-03-09 12:21:51 -0800 Created attachment 131073 Patch 576228 3 131073 mhahnenberg 2012-03-12 10:14:01 -0700 Comment on attachment 131073 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=131073&action=review I'm not an official reviewer, but I figured I'd give you some suggestions for improvement for when an official reviewer does get around to review this :-) > Source/JavaScriptCore/wtf/BumpPointerAllocator.h:114 > + size_t poolSize = std::max(static_cast<size_t>(MINIMUM_BUMP_POOL_SIZE), WTF::pageSize()); What does this change have to do with refactoring CopiedSpace to use WTF::pageSize()? > Source/JavaScriptCore/wtf/StdLibExtras.h:74 > +#define __ROUND_UP_TO_MULTIPLE_OF__ \ Why not make this a function macro? It's kind of confusing to assume variables names from the surrounding scope where the macro is invoked and to have a return statement in the macro itself. Also, naming things with two initial underscores is generally frowned upon because they're typically assumed to be reserved by the compiler. 576229 4 131073 mhahnenberg 2012-03-12 10:14:02 -0700 Comment on attachment 131073 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=131073&action=review I'm not an official reviewer, but I figured I'd give you some suggestions for improvement for when an official reviewer does get around to review this :-) > Source/JavaScriptCore/wtf/BumpPointerAllocator.h:114 > + size_t poolSize = std::max(static_cast<size_t>(MINIMUM_BUMP_POOL_SIZE), WTF::pageSize()); What does this change have to do with refactoring CopiedSpace to use WTF::pageSize()? > Source/JavaScriptCore/wtf/StdLibExtras.h:74 > +#define __ROUND_UP_TO_MULTIPLE_OF__ \ Why not make this a function macro? It's kind of confusing to assume variables names from the surrounding scope where the macro is invoked and to have a return statement in the macro itself. Also, naming things with two initial underscores is generally frowned upon because they're typically assumed to be reserved by the compiler. 576380 5 131073 ggaren 2012-03-12 12:18:47 -0700 Comment on attachment 131073 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=131073&action=review r- based on Mark's comments. > Source/JavaScriptCore/heap/CopiedSpaceInlineMethods.h:174 > + size_t pageMask = ~(WTF::pageSize() - 1); I'd prefer to see a WTF::pageMask() function, rather than one-off code in lots of places. 576779 6 131466 litherum 2012-03-12 17:27:24 -0700 Created attachment 131466 Patch 576790 7 131466 ggaren 2012-03-12 17:35:02 -0700 Comment on attachment 131466 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=131466&action=review > Source/JavaScriptCore/heap/CopiedSpaceInlineMethods.h:174 > + return reinterpret_cast<CopiedBlock*>(reinterpret_cast<size_t>(ptr) & WTF::pageMask()); This function is hot, so I'd prefer not to turn a constant into two out-of-line function calls. I'd suggest making WTF::pageSize() and WTF::pageMask() inline function calls that read globals inside WTF. You can initialize these globals inside WTF::initializeThreading(). > Source/JavaScriptCore/wtf/StdLibExtras.h:76 > +#define ROUND_UP_TO_MULTIPLE_OF(divisor, x) \ > + size_t remainderMask = divisor - 1; \ > + return (x + remainderMask) & ~remainderMask; I'd prefer to see this as an inline function. In fact, you've already written that inline function. So now, all you need is to make you function template call your other function. > Source/JavaScriptCore/wtf/StdLibExtras.h:166 > +inline size_t roundUpToMultipleOf(size_t divisor, size_t x) This function should ASSERT that "divisor" is a power of two. 576794 8 fpizlo 2012-03-12 17:37:19 -0700 (In reply to comment #7) > (From update of attachment 131466 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=131466&action=review > > > Source/JavaScriptCore/heap/CopiedSpaceInlineMethods.h:174 > > + return reinterpret_cast<CopiedBlock*>(reinterpret_cast<size_t>(ptr) & WTF::pageMask()); > > This function is hot, so I'd prefer not to turn a constant into two out-of-line function calls. > > I'd suggest making WTF::pageSize() and WTF::pageMask() inline function calls that read globals inside WTF. You can initialize these globals inside WTF::initializeThreading(). > This is yucky since pageSize() is not known at compile-time. I think we should be using our own fake notion of page size (64KB?) whenever possible and just asserting at run-time that it is both larger than pageSize() and is a (power-of-two) multiple of it. 576820 9 131466 litherum 2012-03-12 17:55:39 -0700 Comment on attachment 131466 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=131466&action=review >>> Source/JavaScriptCore/heap/CopiedSpaceInlineMethods.h:174 >>> + return reinterpret_cast<CopiedBlock*>(reinterpret_cast<size_t>(ptr) & WTF::pageMask()); >> >> This function is hot, so I'd prefer not to turn a constant into two out-of-line function calls. >> >> I'd suggest making WTF::pageSize() and WTF::pageMask() inline function calls that read globals inside WTF. You can initialize these globals inside WTF::initializeThreading(). > > This is yucky since pageSize() is not known at compile-time. > > I think we should be using our own fake notion of page size (64KB?) whenever possible and just asserting at run-time that it is both larger than pageSize() and is a (power-of-two) multiple of it. pageSize() already reads a constant which gets set once: if (!s_pageSize) s_pageSize = systemPageSize(); ... return s_pageSize; Should I continue this strategy and add another size_t to PageBlock.cpp? Or should I initialize the value in initializeThreading() like you said? Moving it to initializeThreading would make pageSize() slightly faster, as it wouldn't have to check to see if the constant is already initialized. Or, should I use a constant page size like Filip Pizlo said? I think using a constant page size doesn't make as much sense as using the system page size, initializing it once, and caching it. I'm happy to do whatever you guys decide on; I'm just getting mixed messages. 576827 10 fpizlo 2012-03-12 18:00:45 -0700 (In reply to comment #9) > (From update of attachment 131466 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=131466&action=review > > >>> Source/JavaScriptCore/heap/CopiedSpaceInlineMethods.h:174 > >>> + return reinterpret_cast<CopiedBlock*>(reinterpret_cast<size_t>(ptr) & WTF::pageMask()); > >> > >> This function is hot, so I'd prefer not to turn a constant into two out-of-line function calls. > >> > >> I'd suggest making WTF::pageSize() and WTF::pageMask() inline function calls that read globals inside WTF. You can initialize these globals inside WTF::initializeThreading(). > > > > This is yucky since pageSize() is not known at compile-time. > > > > I think we should be using our own fake notion of page size (64KB?) whenever possible and just asserting at run-time that it is both larger than pageSize() and is a (power-of-two) multiple of it. > > pageSize() already reads a constant which gets set once: if (!s_pageSize) s_pageSize = systemPageSize(); ... return s_pageSize; > > Should I continue this strategy and add another size_t to PageBlock.cpp? Or should I initialize the value in initializeThreading() like you said? Moving it to initializeThreading would make pageSize() slightly faster, as it wouldn't have to check to see if the constant is already initialized. Or, should I use a constant page size like Filip Pizlo said? I think using a constant page size doesn't make as much sense as using the system page size, initializing it once, and caching it. Apologies, I hadn't read your patch closely enough. This is actually not a hot path - it's the slow case when doing oversize allocation. I think your use of pageSize(), even without making Geoff's suggested changes, is fine. 576829 11 fpizlo 2012-03-12 18:02:09 -0700 (In reply to comment #7) > (From update of attachment 131466 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=131466&action=review > > > Source/JavaScriptCore/heap/CopiedSpaceInlineMethods.h:174 > > + return reinterpret_cast<CopiedBlock*>(reinterpret_cast<size_t>(ptr) & WTF::pageMask()); > > This function is hot, so I'd prefer not to turn a constant into two out-of-line function calls. > > I'd suggest making WTF::pageSize() and WTF::pageMask() inline function calls that read globals inside WTF. You can initialize these globals inside WTF::initializeThreading(). This function should not be hot and even if it was, we'd have already lost. So I think it should be fine to use WTF::pageSize() and WTF::pageMask() without making them inline. That simplifies this patch, and has the nice property that we're not unnecessarily making things inline even though we're not using them in hot paths. Geoff, do you agree? > > > Source/JavaScriptCore/wtf/StdLibExtras.h:76 > > +#define ROUND_UP_TO_MULTIPLE_OF(divisor, x) \ > > + size_t remainderMask = divisor - 1; \ > > + return (x + remainderMask) & ~remainderMask; > > I'd prefer to see this as an inline function. In fact, you've already written that inline function. So now, all you need is to make you function template call your other function. > > > Source/JavaScriptCore/wtf/StdLibExtras.h:166 > > +inline size_t roundUpToMultipleOf(size_t divisor, size_t x) > > This function should ASSERT that "divisor" is a power of two. I concur with Geoff's other two comments. 576886 12 131466 litherum 2012-03-12 18:49:02 -0700 Comment on attachment 131466 Patch View in context: https://bugs.webkit.org/attachment.cgi?id=131466&action=review >>>>>> Source/JavaScriptCore/heap/CopiedSpaceInlineMethods.h:174 >>>>>> + return reinterpret_cast<CopiedBlock*>(reinterpret_cast<size_t>(ptr) & WTF::pageMask()); >>>>> >>>>> This function is hot, so I'd prefer not to turn a constant into two out-of-line function calls. >>>>> >>>>> I'd suggest making WTF::pageSize() and WTF::pageMask() inline function calls that read globals inside WTF. You can initialize these globals inside WTF::initializeThreading(). >>>> >>>> This is yucky since pageSize() is not known at compile-time. >>>> >>>> I think we should be using our own fake notion of page size (64KB?) whenever possible and just asserting at run-time that it is both larger than pageSize() and is a (power-of-two) multiple of it. >>> >>> pageSize() already reads a constant which gets set once: if (!s_pageSize) s_pageSize = systemPageSize(); ... return s_pageSize; >>> >>> Should I continue this strategy and add another size_t to PageBlock.cpp? Or should I initialize the value in initializeThreading() like you said? Moving it to initializeThreading would make pageSize() slightly faster, as it wouldn't have to check to see if the constant is already initialized. Or, should I use a constant page size like Filip Pizlo said? I think using a constant page size doesn't make as much sense as using the system page size, initializing it once, and caching it. >>> >>> I'm happy to do whatever you guys decide on; I'm just getting mixed messages. >> >> Apologies, I hadn't read your patch closely enough. This is actually not a hot path - it's the slow case when doing oversize allocation. >> >> I think your use of pageSize(), even without making Geoff's suggested changes, is fine. > > This function should not be hot and even if it was, we'd have already lost. So I think it should be fine to use WTF::pageSize() and WTF::pageMask() without making them inline. That simplifies this patch, and has the nice property that we're not unnecessarily making things inline even though we're not using them in hot paths. > > Geoff, do you agree? While inlining pageMask() and pageSize(), I'm coming across some pain-points: - initializeThreading will be doing something that is completely unrelated to threading - initializeThreading is system-specific, but all the systems need to set WTF::s_pageSize and WTF::s_pageMask. Doing this properly (having a generic initializeThreading routine that calls the system-specific one) seems like a very major change for this CL - inlining pageSize and pageMask means I have to move their definitions to the header (PageBlock.h). Because I'd be moving systemPageSize out of PageBlock.cpp (to live next to initializeThreading), PageBlock.cpp won't have any functions defined in it. I'd then be deleting PageBlock.cpp because it would be empty. Does this sounds reasonable? I wanted to know if I should proceed before changing how initializeThreading is called and deleting an entire file. 576930 13 fpizlo 2012-03-12 19:37:08 -0700 As I mentioned in my previous two comments, inlining WTF::pageSize is unnecessary and probably unwise. (In reply to comment #12) > (From update of attachment 131466 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=131466&action=review > > >>>>>> Source/JavaScriptCore/heap/CopiedSpaceInlineMethods.h:174 > >>>>>> + return reinterpret_cast<CopiedBlock*>(reinterpret_cast<size_t>(ptr) & WTF::pageMask()); > >>>>> > >>>>> This function is hot, so I'd prefer not to turn a constant into two out-of-line function calls. > >>>>> > >>>>> I'd suggest making WTF::pageSize() and WTF::pageMask() inline function calls that read globals inside WTF. You can initialize these globals inside WTF::initializeThreading(). > >>>> > >>>> This is yucky since pageSize() is not known at compile-time. > >>>> > >>>> I think we should be using our own fake notion of page size (64KB?) whenever possible and just asserting at run-time that it is both larger than pageSize() and is a (power-of-two) multiple of it. > >>> > >>> pageSize() already reads a constant which gets set once: if (!s_pageSize) s_pageSize = systemPageSize(); ... return s_pageSize; > >>> > >>> Should I continue this strategy and add another size_t to PageBlock.cpp? Or should I initialize the value in initializeThreading() like you said? Moving it to initializeThreading would make pageSize() slightly faster, as it wouldn't have to check to see if the constant is already initialized. Or, should I use a constant page size like Filip Pizlo said? I think using a constant page size doesn't make as much sense as using the system page size, initializing it once, and caching it. > >>> > >>> I'm happy to do whatever you guys decide on; I'm just getting mixed messages. > >> > >> Apologies, I hadn't read your patch closely enough. This is actually not a hot path - it's the slow case when doing oversize allocation. > >> > >> I think your use of pageSize(), even without making Geoff's suggested changes, is fine. > > > > This function should not be hot and even if it was, we'd have already lost. So I think it should be fine to use WTF::pageSize() and WTF::pageMask() without making them inline. That simplifies this patch, and has the nice property that we're not unnecessarily making things inline even though we're not using them in hot paths. > > > > Geoff, do you agree? > > While inlining pageMask() and pageSize(), I'm coming across some pain-points: > > - initializeThreading will be doing something that is completely unrelated to threading > - initializeThreading is system-specific, but all the systems need to set WTF::s_pageSize and WTF::s_pageMask. Doing this properly (having a generic initializeThreading routine that calls the system-specific one) seems like a very major change for this CL > - inlining pageSize and pageMask means I have to move their definitions to the header (PageBlock.h). Because I'd be moving systemPageSize out of PageBlock.cpp (to live next to initializeThreading), PageBlock.cpp won't have any functions defined in it. I'd then be deleting PageBlock.cpp because it would be empty. > > Does this sounds reasonable? I wanted to know if I should proceed before changing how initializeThreading is called and deleting an entire file. 577539 14 131667 litherum 2012-03-13 10:54:33 -0700 Created attachment 131667 Patch 579743 15 ggaren 2012-03-15 14:30:30 -0700 > This function should not be hot and even if it was, we'd have already lost. Agreed -- as of now. As such, I'll say "r+". If we aim to use these functions throughout the system, in place of constants, we'll eventually run into some hot code paths, and we'll need to revisit the inlining question. I'm not convinced that there's any fundamental barrier to inlining here, but I am convinced that we can cross that bridge when we come to it. 579744 16 131667 ggaren 2012-03-15 14:30:41 -0700 Comment on attachment 131667 Patch r=me 579844 17 131667 webkit.review.bot 2012-03-15 16:04:54 -0700 Comment on attachment 131667 Patch Clearing flags on attachment: 131667 Committed r110902: <http://trac.webkit.org/changeset/110902> 579845 18 webkit.review.bot 2012-03-15 16:04:59 -0700 All reviewed patches have been landed. Closing bug. 131073 2012-03-09 12:21:51 -0800 2012-03-12 17:27:19 -0700 Patch bug-80615-20120309122212.patch text/plain 5237 litherum SW5kZXg6IFNvdXJjZS9KYXZhU2NyaXB0Q29yZS9DaGFuZ2VMb2cKPT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gU291 cmNlL0phdmFTY3JpcHRDb3JlL0NoYW5nZUxvZwkocmV2aXNpb24gMTEwMzE2KQorKysgU291cmNl L0phdmFTY3JpcHRDb3JlL0NoYW5nZUxvZwkod29ya2luZyBjb3B5KQpAQCAtMSwzICsxLDIxIEBA CisyMDEyLTAzLTA5ICBNeWxlcyBNYXhmaWVsZCAgPG1tYXhmaWVsZEBnb29nbGUuY29tPgorCisg ICAgICAgIENvcGllZFNwYWNlOjp0cnlBbGxvY2F0ZU92ZXJzaXplIGFzc3VtZXMgc3lzdGVtIHBh Z2Ugc2l6ZQorICAgICAgICBodHRwczovL2J1Z3Mud2Via2l0Lm9yZy9zaG93X2J1Zy5jZ2k/aWQ9 ODA2MTUKKworICAgICAgICBSZXZpZXdlZCBieSBOT0JPRFkgKE9PUFMhKS4KKworICAgICAgICAq IGhlYXAvQ29waWVkU3BhY2UuY3BwOgorICAgICAgICAoSlNDOjpDb3BpZWRTcGFjZTo6dHJ5QWxs b2NhdGVPdmVyc2l6ZSk6CisgICAgICAgICogaGVhcC9Db3BpZWRTcGFjZS5oOgorICAgICAgICAo Q29waWVkU3BhY2UpOgorICAgICAgICAqIGhlYXAvQ29waWVkU3BhY2VJbmxpbmVNZXRob2RzLmg6 CisgICAgICAgIChKU0M6OkNvcGllZFNwYWNlOjpvdmVyc2l6ZUJsb2NrRm9yKToKKyAgICAgICAg KiB3dGYvQnVtcFBvaW50ZXJBbGxvY2F0b3IuaDoKKyAgICAgICAgKFdURjo6QnVtcFBvaW50ZXJQ b29sOjpjcmVhdGUpOgorICAgICAgICAqIHd0Zi9TdGRMaWJFeHRyYXMuaDoKKyAgICAgICAgKFdU Rjo6cm91bmRVcFRvTXVsdGlwbGVPZik6CisKIDIwMTItMDMtMDkgIEpvbiBMZWUgIDxqb25sZWVA YXBwbGUuY29tPgogCiAgICAgICAgIEFkZCBzdXBwb3J0IGZvciBFTkFCTEUoTEVHQUNZX05PVElG SUNBVElPTlMpCkluZGV4OiBTb3VyY2UvSmF2YVNjcmlwdENvcmUvaGVhcC9Db3BpZWRTcGFjZS5j cHAKPT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PQotLS0gU291cmNlL0phdmFTY3JpcHRDb3JlL2hlYXAvQ29waWVkU3BhY2Uu Y3BwCShyZXZpc2lvbiAxMTAxOTcpCisrKyBTb3VyY2UvSmF2YVNjcmlwdENvcmUvaGVhcC9Db3Bp ZWRTcGFjZS5jcHAJKHdvcmtpbmcgY29weSkKQEAgLTcxLDggKzcxLDggQEAgQ2hlY2tlZEJvb2xl YW4gQ29waWVkU3BhY2U6OnRyeUFsbG9jYXRlTwogewogICAgIEFTU0VSVChpc092ZXJzaXplKGJ5 dGVzKSk7CiAgICAgCi0gICAgc2l6ZV90IGJsb2NrU2l6ZSA9IFdURjo6cm91bmRVcFRvTXVsdGlw bGVPZjxzX3BhZ2VTaXplPihzaXplb2YoQ29waWVkQmxvY2spICsgYnl0ZXMpOwotICAgIFBhZ2VB bGxvY2F0aW9uQWxpZ25lZCBhbGxvY2F0aW9uID0gUGFnZUFsbG9jYXRpb25BbGlnbmVkOjphbGxv Y2F0ZShibG9ja1NpemUsIHNfcGFnZVNpemUsIE9TQWxsb2NhdG9yOjpKU0dDSGVhcFBhZ2VzKTsK KyAgICBzaXplX3QgYmxvY2tTaXplID0gV1RGOjpyb3VuZFVwVG9NdWx0aXBsZU9mKFdURjo6cGFn ZVNpemUoKSwgc2l6ZW9mKENvcGllZEJsb2NrKSArIGJ5dGVzKTsKKyAgICBQYWdlQWxsb2NhdGlv bkFsaWduZWQgYWxsb2NhdGlvbiA9IFBhZ2VBbGxvY2F0aW9uQWxpZ25lZDo6YWxsb2NhdGUoYmxv Y2tTaXplLCBXVEY6OnBhZ2VTaXplKCksIE9TQWxsb2NhdG9yOjpKU0dDSGVhcFBhZ2VzKTsKICAg ICBpZiAoIXN0YXRpY19jYXN0PGJvb2w+KGFsbG9jYXRpb24pKSB7CiAgICAgICAgICpvdXRQdHIg PSAwOwogICAgICAgICByZXR1cm4gZmFsc2U7CkluZGV4OiBTb3VyY2UvSmF2YVNjcmlwdENvcmUv aGVhcC9Db3BpZWRTcGFjZS5oCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIFNvdXJjZS9KYXZhU2NyaXB0Q29yZS9o ZWFwL0NvcGllZFNwYWNlLmgJKHJldmlzaW9uIDExMDE5NykKKysrIFNvdXJjZS9KYXZhU2NyaXB0 Q29yZS9oZWFwL0NvcGllZFNwYWNlLmgJKHdvcmtpbmcgY29weSkKQEAgLTM1LDYgKzM1LDcgQEAK ICNpbmNsdWRlIDx3dGYvSGFzaFNldC5oPgogI2luY2x1ZGUgPHd0Zi9PU0FsbG9jYXRvci5oPgog I2luY2x1ZGUgPHd0Zi9QYWdlQWxsb2NhdGlvbkFsaWduZWQuaD4KKyNpbmNsdWRlIDx3dGYvUGFn ZUJsb2NrLmg+CiAjaW5jbHVkZSA8d3RmL1N0ZExpYkV4dHJhcy5oPgogI2luY2x1ZGUgPHd0Zi9U aHJlYWRpbmdQcmltaXRpdmVzLmg+CiAKQEAgLTExNiw4ICsxMTcsNiBAQCBwcml2YXRlOgogICAg IHNpemVfdCBtX251bWJlck9mTG9hbmVkQmxvY2tzOwogCiAgICAgc3RhdGljIGNvbnN0IHNpemVf dCBzX21heEFsbG9jYXRpb25TaXplID0gMzIgKiBLQjsKLSAgICBzdGF0aWMgY29uc3Qgc2l6ZV90 IHNfcGFnZVNpemUgPSA0ICogS0I7Ci0gICAgc3RhdGljIGNvbnN0IHNpemVfdCBzX3BhZ2VNYXNr ID0gfihzX3BhZ2VTaXplIC0gMSk7CiAgICAgc3RhdGljIGNvbnN0IHNpemVfdCBzX2luaXRpYWxC bG9ja051bSA9IDE2OwogICAgIHN0YXRpYyBjb25zdCBzaXplX3Qgc19ibG9ja01hc2sgPSB+KEhl YXBCbG9jazo6c19ibG9ja1NpemUgLSAxKTsKIH07CkluZGV4OiBTb3VyY2UvSmF2YVNjcmlwdENv cmUvaGVhcC9Db3BpZWRTcGFjZUlubGluZU1ldGhvZHMuaAo9PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBTb3VyY2Uv SmF2YVNjcmlwdENvcmUvaGVhcC9Db3BpZWRTcGFjZUlubGluZU1ldGhvZHMuaAkocmV2aXNpb24g MTEwMTk3KQorKysgU291cmNlL0phdmFTY3JpcHRDb3JlL2hlYXAvQ29waWVkU3BhY2VJbmxpbmVN ZXRob2RzLmgJKHdvcmtpbmcgY29weSkKQEAgLTE3MSw3ICsxNzEsOCBAQCBpbmxpbmUgYm9vbCBD b3BpZWRTcGFjZTo6aXNQaW5uZWQodm9pZCogCiAKIGlubGluZSBDb3BpZWRCbG9jayogQ29waWVk U3BhY2U6Om92ZXJzaXplQmxvY2tGb3Iodm9pZCogcHRyKQogewotICAgIHJldHVybiByZWludGVy cHJldF9jYXN0PENvcGllZEJsb2NrKj4ocmVpbnRlcnByZXRfY2FzdDxzaXplX3Q+KHB0cikgJiBz X3BhZ2VNYXNrKTsKKyAgICBzaXplX3QgcGFnZU1hc2sgPSB+KFdURjo6cGFnZVNpemUoKSAtIDEp OworICAgIHJldHVybiByZWludGVycHJldF9jYXN0PENvcGllZEJsb2NrKj4ocmVpbnRlcnByZXRf Y2FzdDxzaXplX3Q+KHB0cikgJiBwYWdlTWFzayk7CiB9CiAKIGlubGluZSBDb3BpZWRCbG9jayog Q29waWVkU3BhY2U6OmJsb2NrRm9yKHZvaWQqIHB0cikKSW5kZXg6IFNvdXJjZS9KYXZhU2NyaXB0 Q29yZS93dGYvQnVtcFBvaW50ZXJBbGxvY2F0b3IuaAo9PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBTb3VyY2UvSmF2 YVNjcmlwdENvcmUvd3RmL0J1bXBQb2ludGVyQWxsb2NhdG9yLmgJKHJldmlzaW9uIDExMDE5NykK KysrIFNvdXJjZS9KYXZhU2NyaXB0Q29yZS93dGYvQnVtcFBvaW50ZXJBbGxvY2F0b3IuaAkod29y a2luZyBjb3B5KQpAQCAtMjYsNiArMjYsNyBAQAogI2lmbmRlZiBCdW1wUG9pbnRlckFsbG9jYXRv cl9oCiAjZGVmaW5lIEJ1bXBQb2ludGVyQWxsb2NhdG9yX2gKIAorI2luY2x1ZGUgPGFsZ29yaXRo bT4KICNpbmNsdWRlIDx3dGYvUGFnZUFsbG9jYXRpb24uaD4KIAogbmFtZXNwYWNlIFdURiB7CkBA IC0xMTAsNyArMTExLDcgQEAgcHJpdmF0ZToKICAgICAgICAgaWYgKG1pbmltdW1DYXBhY2l0eSA8 IHNpemVvZihCdW1wUG9pbnRlclBvb2wpKQogICAgICAgICAgICAgcmV0dXJuIDA7CiAKLSAgICAg ICAgc2l6ZV90IHBvb2xTaXplID0gTUlOSU1VTV9CVU1QX1BPT0xfU0laRTsKKyAgICAgICAgc2l6 ZV90IHBvb2xTaXplID0gc3RkOjptYXgoc3RhdGljX2Nhc3Q8c2l6ZV90PihNSU5JTVVNX0JVTVBf UE9PTF9TSVpFKSwgV1RGOjpwYWdlU2l6ZSgpKTsKICAgICAgICAgd2hpbGUgKHBvb2xTaXplIDwg bWluaW11bUNhcGFjaXR5KSB7CiAgICAgICAgICAgICBwb29sU2l6ZSA8PD0gMTsKICAgICAgICAg ICAgIC8vIFRoZSBmb2xsb3dpbmcgaWYgY2hlY2sgcmVsaWVzIG9uIE1JTklNVU1fQlVNUF9QT09M X1NJWkUgYmVpbmcgYSBwb3dlciBvZiAyIQpJbmRleDogU291cmNlL0phdmFTY3JpcHRDb3JlL3d0 Zi9TdGRMaWJFeHRyYXMuaAo9PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBTb3VyY2UvSmF2YVNjcmlwdENvcmUvd3Rm L1N0ZExpYkV4dHJhcy5oCShyZXZpc2lvbiAxMTAxOTcpCisrKyBTb3VyY2UvSmF2YVNjcmlwdENv cmUvd3RmL1N0ZExpYkV4dHJhcy5oCSh3b3JraW5nIGNvcHkpCkBAIC03MSw2ICs3MSwxMCBAQAog I2RlZmluZSBTVFJJTkdJWkUoZXhwKSAjZXhwCiAjZGVmaW5lIFNUUklOR0laRV9WQUxVRV9PRihl eHApIFNUUklOR0laRShleHApCiAKKyNkZWZpbmUgX19ST1VORF9VUF9UT19NVUxUSVBMRV9PRl9f IFwKKyAgICBzaXplX3QgcmVtYWluZGVyTWFzayA9IGRpdmlzb3IgLSAxOyBcCisgICAgcmV0dXJu ICh4ICsgcmVtYWluZGVyTWFzaykgJiB+cmVtYWluZGVyTWFzazsKKwogLyoKICAqIFRoZSByZWlu dGVycHJldF9jYXN0PFR5cGUxKj4oW3BvaW50ZXIgdG8gVHlwZTJdKSBleHByZXNzaW9ucyAtIHdo ZXJlCiAgKiBzaXplb2YoVHlwZTEpID4gc2l6ZW9mKFR5cGUyKSAtIGNhdXNlIHRoZSBmb2xsb3dp bmcgd2FybmluZyBvbiBBUk0gd2l0aCBHQ0M6CkBAIC0xNTcsOSArMTYxLDExIEBAIHRlbXBsYXRl PHR5cGVuYW1lIFQsIHNpemVfdCBTaXplPiBjaGFyICgKIHRlbXBsYXRlPHNpemVfdCBkaXZpc29y PiBpbmxpbmUgc2l6ZV90IHJvdW5kVXBUb011bHRpcGxlT2Yoc2l6ZV90IHgpCiB7CiAgICAgQ09N UElMRV9BU1NFUlQoZGl2aXNvciAmJiAhKGRpdmlzb3IgJiAoZGl2aXNvciAtIDEpKSwgZGl2aXNv cl9pc19hX3Bvd2VyX29mX3R3byk7Ci0KLSAgICBzaXplX3QgcmVtYWluZGVyTWFzayA9IGRpdmlz b3IgLSAxOwotICAgIHJldHVybiAoeCArIHJlbWFpbmRlck1hc2spICYgfnJlbWFpbmRlck1hc2s7 CisgICAgX19ST1VORF9VUF9UT19NVUxUSVBMRV9PRl9fCit9CitpbmxpbmUgc2l6ZV90IHJvdW5k VXBUb011bHRpcGxlT2Yoc2l6ZV90IGRpdmlzb3IsIHNpemVfdCB4KQoreworICAgIF9fUk9VTkRf VVBfVE9fTVVMVElQTEVfT0ZfXwogfQogCiBlbnVtIEJpbmFyeVNlYXJjaE1vZGUgewo= 131466 2012-03-12 17:27:24 -0700 2012-03-13 10:54:15 -0700 Patch bug-80615-20120312172736.patch text/plain 5419 litherum SW5kZXg6IFNvdXJjZS9KYXZhU2NyaXB0Q29yZS9DaGFuZ2VMb2cKPT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gU291 cmNlL0phdmFTY3JpcHRDb3JlL0NoYW5nZUxvZwkocmV2aXNpb24gMTEwMzU5KQorKysgU291cmNl L0phdmFTY3JpcHRDb3JlL0NoYW5nZUxvZwkod29ya2luZyBjb3B5KQpAQCAtMSwzICsxLDIxIEBA CisyMDEyLTAzLTA5ICBNeWxlcyBNYXhmaWVsZCAgPG1tYXhmaWVsZEBnb29nbGUuY29tPgorCisg ICAgICAgIENvcGllZFNwYWNlOjp0cnlBbGxvY2F0ZU92ZXJzaXplIGFzc3VtZXMgc3lzdGVtIHBh Z2Ugc2l6ZQorICAgICAgICBodHRwczovL2J1Z3Mud2Via2l0Lm9yZy9zaG93X2J1Zy5jZ2k/aWQ9 ODA2MTUKKworICAgICAgICBSZXZpZXdlZCBieSBOT0JPRFkgKE9PUFMhKS4KKworICAgICAgICAq IGhlYXAvQ29waWVkU3BhY2UuY3BwOgorICAgICAgICAoSlNDOjpDb3BpZWRTcGFjZTo6dHJ5QWxs b2NhdGVPdmVyc2l6ZSk6CisgICAgICAgICogaGVhcC9Db3BpZWRTcGFjZS5oOgorICAgICAgICAo Q29waWVkU3BhY2UpOgorICAgICAgICAqIGhlYXAvQ29waWVkU3BhY2VJbmxpbmVNZXRob2RzLmg6 CisgICAgICAgIChKU0M6OkNvcGllZFNwYWNlOjpvdmVyc2l6ZUJsb2NrRm9yKToKKyAgICAgICAg KiB3dGYvQnVtcFBvaW50ZXJBbGxvY2F0b3IuaDoKKyAgICAgICAgKFdURjo6QnVtcFBvaW50ZXJQ b29sOjpjcmVhdGUpOgorICAgICAgICAqIHd0Zi9TdGRMaWJFeHRyYXMuaDoKKyAgICAgICAgKFdU Rjo6cm91bmRVcFRvTXVsdGlwbGVPZik6CisKIDIwMTItMDMtMDkgIEdhdmluIEJhcnJhY2xvdWdo ICA8YmFycmFjbG91Z2hAYXBwbGUuY29tPgogCiAgICAgICAgIEFycmF5LnByb3RvdHlwZS50b0xv Y2FsZVN0cmluZyB2aXNpdHMgZWxlbWVudHMgaW4gd3Jvbmcgb3JkZXIgdW5kZXIgY2VydGFpbiBj b25kaXRpb25zCkluZGV4OiBTb3VyY2UvSmF2YVNjcmlwdENvcmUvaGVhcC9Db3BpZWRTcGFjZS5j cHAKPT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PQotLS0gU291cmNlL0phdmFTY3JpcHRDb3JlL2hlYXAvQ29waWVkU3BhY2Uu Y3BwCShyZXZpc2lvbiAxMTAzNTkpCisrKyBTb3VyY2UvSmF2YVNjcmlwdENvcmUvaGVhcC9Db3Bp ZWRTcGFjZS5jcHAJKHdvcmtpbmcgY29weSkKQEAgLTcxLDggKzcxLDggQEAgQ2hlY2tlZEJvb2xl YW4gQ29waWVkU3BhY2U6OnRyeUFsbG9jYXRlTwogewogICAgIEFTU0VSVChpc092ZXJzaXplKGJ5 dGVzKSk7CiAgICAgCi0gICAgc2l6ZV90IGJsb2NrU2l6ZSA9IFdURjo6cm91bmRVcFRvTXVsdGlw bGVPZjxzX3BhZ2VTaXplPihzaXplb2YoQ29waWVkQmxvY2spICsgYnl0ZXMpOwotICAgIFBhZ2VB bGxvY2F0aW9uQWxpZ25lZCBhbGxvY2F0aW9uID0gUGFnZUFsbG9jYXRpb25BbGlnbmVkOjphbGxv Y2F0ZShibG9ja1NpemUsIHNfcGFnZVNpemUsIE9TQWxsb2NhdG9yOjpKU0dDSGVhcFBhZ2VzKTsK KyAgICBzaXplX3QgYmxvY2tTaXplID0gV1RGOjpyb3VuZFVwVG9NdWx0aXBsZU9mKFdURjo6cGFn ZVNpemUoKSwgc2l6ZW9mKENvcGllZEJsb2NrKSArIGJ5dGVzKTsKKyAgICBQYWdlQWxsb2NhdGlv bkFsaWduZWQgYWxsb2NhdGlvbiA9IFBhZ2VBbGxvY2F0aW9uQWxpZ25lZDo6YWxsb2NhdGUoYmxv Y2tTaXplLCBXVEY6OnBhZ2VTaXplKCksIE9TQWxsb2NhdG9yOjpKU0dDSGVhcFBhZ2VzKTsKICAg ICBpZiAoIXN0YXRpY19jYXN0PGJvb2w+KGFsbG9jYXRpb24pKSB7CiAgICAgICAgICpvdXRQdHIg PSAwOwogICAgICAgICByZXR1cm4gZmFsc2U7CkluZGV4OiBTb3VyY2UvSmF2YVNjcmlwdENvcmUv aGVhcC9Db3BpZWRTcGFjZS5oCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIFNvdXJjZS9KYXZhU2NyaXB0Q29yZS9o ZWFwL0NvcGllZFNwYWNlLmgJKHJldmlzaW9uIDExMDM1OSkKKysrIFNvdXJjZS9KYXZhU2NyaXB0 Q29yZS9oZWFwL0NvcGllZFNwYWNlLmgJKHdvcmtpbmcgY29weSkKQEAgLTM1LDYgKzM1LDcgQEAK ICNpbmNsdWRlIDx3dGYvSGFzaFNldC5oPgogI2luY2x1ZGUgPHd0Zi9PU0FsbG9jYXRvci5oPgog I2luY2x1ZGUgPHd0Zi9QYWdlQWxsb2NhdGlvbkFsaWduZWQuaD4KKyNpbmNsdWRlIDx3dGYvUGFn ZUJsb2NrLmg+CiAjaW5jbHVkZSA8d3RmL1N0ZExpYkV4dHJhcy5oPgogI2luY2x1ZGUgPHd0Zi9U aHJlYWRpbmdQcmltaXRpdmVzLmg+CiAKQEAgLTExNiw4ICsxMTcsNiBAQCBwcml2YXRlOgogICAg IHNpemVfdCBtX251bWJlck9mTG9hbmVkQmxvY2tzOwogCiAgICAgc3RhdGljIGNvbnN0IHNpemVf dCBzX21heEFsbG9jYXRpb25TaXplID0gMzIgKiBLQjsKLSAgICBzdGF0aWMgY29uc3Qgc2l6ZV90 IHNfcGFnZVNpemUgPSA0ICogS0I7Ci0gICAgc3RhdGljIGNvbnN0IHNpemVfdCBzX3BhZ2VNYXNr ID0gfihzX3BhZ2VTaXplIC0gMSk7CiAgICAgc3RhdGljIGNvbnN0IHNpemVfdCBzX2luaXRpYWxC bG9ja051bSA9IDE2OwogICAgIHN0YXRpYyBjb25zdCBzaXplX3Qgc19ibG9ja01hc2sgPSB+KEhl YXBCbG9jazo6c19ibG9ja1NpemUgLSAxKTsKIH07CkluZGV4OiBTb3VyY2UvSmF2YVNjcmlwdENv cmUvaGVhcC9Db3BpZWRTcGFjZUlubGluZU1ldGhvZHMuaAo9PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBTb3VyY2Uv SmF2YVNjcmlwdENvcmUvaGVhcC9Db3BpZWRTcGFjZUlubGluZU1ldGhvZHMuaAkocmV2aXNpb24g MTEwMzU5KQorKysgU291cmNlL0phdmFTY3JpcHRDb3JlL2hlYXAvQ29waWVkU3BhY2VJbmxpbmVN ZXRob2RzLmgJKHdvcmtpbmcgY29weSkKQEAgLTE3MSw3ICsxNzEsNyBAQCBpbmxpbmUgYm9vbCBD b3BpZWRTcGFjZTo6aXNQaW5uZWQodm9pZCogCiAKIGlubGluZSBDb3BpZWRCbG9jayogQ29waWVk U3BhY2U6Om92ZXJzaXplQmxvY2tGb3Iodm9pZCogcHRyKQogewotICAgIHJldHVybiByZWludGVy cHJldF9jYXN0PENvcGllZEJsb2NrKj4ocmVpbnRlcnByZXRfY2FzdDxzaXplX3Q+KHB0cikgJiBz X3BhZ2VNYXNrKTsKKyAgICByZXR1cm4gcmVpbnRlcnByZXRfY2FzdDxDb3BpZWRCbG9jayo+KHJl aW50ZXJwcmV0X2Nhc3Q8c2l6ZV90PihwdHIpICYgV1RGOjpwYWdlTWFzaygpKTsKIH0KIAogaW5s aW5lIENvcGllZEJsb2NrKiBDb3BpZWRTcGFjZTo6YmxvY2tGb3Iodm9pZCogcHRyKQpJbmRleDog U291cmNlL0phdmFTY3JpcHRDb3JlL3d0Zi9QYWdlQmxvY2suY3BwCj09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIFNv dXJjZS9KYXZhU2NyaXB0Q29yZS93dGYvUGFnZUJsb2NrLmNwcAkocmV2aXNpb24gMTEwMzU5KQor KysgU291cmNlL0phdmFTY3JpcHRDb3JlL3d0Zi9QYWdlQmxvY2suY3BwCSh3b3JraW5nIGNvcHkp CkBAIC02Nyw0ICs2Nyw5IEBAIHNpemVfdCBwYWdlU2l6ZSgpCiAgICAgcmV0dXJuIHNfcGFnZVNp emU7CiB9CiAKK3NpemVfdCBwYWdlTWFzaygpCit7CisgICAgcmV0dXJuIH4ocGFnZVNpemUoKSAt IDEpOworfQorCiB9IC8vIG5hbWVzcGFjZSBXVEYKSW5kZXg6IFNvdXJjZS9KYXZhU2NyaXB0Q29y ZS93dGYvUGFnZUJsb2NrLmgKPT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gU291cmNlL0phdmFTY3JpcHRDb3JlL3d0 Zi9QYWdlQmxvY2suaAkocmV2aXNpb24gMTEwMzU5KQorKysgU291cmNlL0phdmFTY3JpcHRDb3Jl L3d0Zi9QYWdlQmxvY2suaAkod29ya2luZyBjb3B5KQpAQCAtMjksNiArMjksNyBAQAogbmFtZXNw YWNlIFdURiB7CiAKIFdURl9FWFBPUlRfUFJJVkFURSBzaXplX3QgcGFnZVNpemUoKTsKK1dURl9F WFBPUlRfUFJJVkFURSBzaXplX3QgcGFnZU1hc2soKTsKIGlubGluZSBib29sIGlzUGFnZUFsaWdu ZWQodm9pZCogYWRkcmVzcykgeyByZXR1cm4gIShyZWludGVycHJldF9jYXN0PGludHB0cl90Pihh ZGRyZXNzKSAmIChwYWdlU2l6ZSgpIC0gMSkpOyB9CiBpbmxpbmUgYm9vbCBpc1BhZ2VBbGlnbmVk KHNpemVfdCBzaXplKSB7IHJldHVybiAhKHNpemUgJiAocGFnZVNpemUoKSAtIDEpKTsgfQogaW5s aW5lIGJvb2wgaXNQb3dlck9mVHdvKHNpemVfdCBzaXplKSB7IHJldHVybiAhKHNpemUgJiAoc2l6 ZSAtIDEpKTsgfQpJbmRleDogU291cmNlL0phdmFTY3JpcHRDb3JlL3d0Zi9TdGRMaWJFeHRyYXMu aAo9PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09Ci0tLSBTb3VyY2UvSmF2YVNjcmlwdENvcmUvd3RmL1N0ZExpYkV4dHJhcy5o CShyZXZpc2lvbiAxMTAzNTkpCisrKyBTb3VyY2UvSmF2YVNjcmlwdENvcmUvd3RmL1N0ZExpYkV4 dHJhcy5oCSh3b3JraW5nIGNvcHkpCkBAIC03MSw2ICs3MSwxMCBAQAogI2RlZmluZSBTVFJJTkdJ WkUoZXhwKSAjZXhwCiAjZGVmaW5lIFNUUklOR0laRV9WQUxVRV9PRihleHApIFNUUklOR0laRShl eHApCiAKKyNkZWZpbmUgUk9VTkRfVVBfVE9fTVVMVElQTEVfT0YoZGl2aXNvciwgeCkgXAorICAg IHNpemVfdCByZW1haW5kZXJNYXNrID0gZGl2aXNvciAtIDE7IFwKKyAgICByZXR1cm4gKHggKyBy ZW1haW5kZXJNYXNrKSAmIH5yZW1haW5kZXJNYXNrOworCiAvKgogICogVGhlIHJlaW50ZXJwcmV0 X2Nhc3Q8VHlwZTEqPihbcG9pbnRlciB0byBUeXBlMl0pIGV4cHJlc3Npb25zIC0gd2hlcmUKICAq IHNpemVvZihUeXBlMSkgPiBzaXplb2YoVHlwZTIpIC0gY2F1c2UgdGhlIGZvbGxvd2luZyB3YXJu aW5nIG9uIEFSTSB3aXRoIEdDQzoKQEAgLTE1Nyw5ICsxNjEsMTEgQEAgdGVtcGxhdGU8dHlwZW5h bWUgVCwgc2l6ZV90IFNpemU+IGNoYXIgKAogdGVtcGxhdGU8c2l6ZV90IGRpdmlzb3I+IGlubGlu ZSBzaXplX3Qgcm91bmRVcFRvTXVsdGlwbGVPZihzaXplX3QgeCkKIHsKICAgICBDT01QSUxFX0FT U0VSVChkaXZpc29yICYmICEoZGl2aXNvciAmIChkaXZpc29yIC0gMSkpLCBkaXZpc29yX2lzX2Ff cG93ZXJfb2ZfdHdvKTsKLQotICAgIHNpemVfdCByZW1haW5kZXJNYXNrID0gZGl2aXNvciAtIDE7 Ci0gICAgcmV0dXJuICh4ICsgcmVtYWluZGVyTWFzaykgJiB+cmVtYWluZGVyTWFzazsKKyAgICBS T1VORF9VUF9UT19NVUxUSVBMRV9PRihkaXZpc29yLCB4KTsKK30KK2lubGluZSBzaXplX3Qgcm91 bmRVcFRvTXVsdGlwbGVPZihzaXplX3QgZGl2aXNvciwgc2l6ZV90IHgpCit7CisgICAgUk9VTkRf VVBfVE9fTVVMVElQTEVfT0YoZGl2aXNvciwgeCk7CiB9CiAKIGVudW0gQmluYXJ5U2VhcmNoTW9k ZSB7Cg== 131667 2012-03-13 10:54:33 -0700 2012-03-15 16:04:54 -0700 Patch bug-80615-20120313105432.patch text/plain 5445 litherum SW5kZXg6IFNvdXJjZS9KYXZhU2NyaXB0Q29yZS9DaGFuZ2VMb2cKPT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQotLS0gU291 cmNlL0phdmFTY3JpcHRDb3JlL0NoYW5nZUxvZwkocmV2aXNpb24gMTEwMzU5KQorKysgU291cmNl L0phdmFTY3JpcHRDb3JlL0NoYW5nZUxvZwkod29ya2luZyBjb3B5KQpAQCAtMSwzICsxLDIxIEBA CisyMDEyLTAzLTA5ICBNeWxlcyBNYXhmaWVsZCAgPG1tYXhmaWVsZEBnb29nbGUuY29tPgorCisg ICAgICAgIENvcGllZFNwYWNlOjp0cnlBbGxvY2F0ZU92ZXJzaXplIGFzc3VtZXMgc3lzdGVtIHBh Z2Ugc2l6ZQorICAgICAgICBodHRwczovL2J1Z3Mud2Via2l0Lm9yZy9zaG93X2J1Zy5jZ2k/aWQ9 ODA2MTUKKworICAgICAgICBSZXZpZXdlZCBieSBOT0JPRFkgKE9PUFMhKS4KKworICAgICAgICAq IGhlYXAvQ29waWVkU3BhY2UuY3BwOgorICAgICAgICAoSlNDOjpDb3BpZWRTcGFjZTo6dHJ5QWxs b2NhdGVPdmVyc2l6ZSk6CisgICAgICAgICogaGVhcC9Db3BpZWRTcGFjZS5oOgorICAgICAgICAo Q29waWVkU3BhY2UpOgorICAgICAgICAqIGhlYXAvQ29waWVkU3BhY2VJbmxpbmVNZXRob2RzLmg6 CisgICAgICAgIChKU0M6OkNvcGllZFNwYWNlOjpvdmVyc2l6ZUJsb2NrRm9yKToKKyAgICAgICAg KiB3dGYvQnVtcFBvaW50ZXJBbGxvY2F0b3IuaDoKKyAgICAgICAgKFdURjo6QnVtcFBvaW50ZXJQ b29sOjpjcmVhdGUpOgorICAgICAgICAqIHd0Zi9TdGRMaWJFeHRyYXMuaDoKKyAgICAgICAgKFdU Rjo6cm91bmRVcFRvTXVsdGlwbGVPZik6CisKIDIwMTItMDMtMDkgIEdhdmluIEJhcnJhY2xvdWdo ICA8YmFycmFjbG91Z2hAYXBwbGUuY29tPgogCiAgICAgICAgIEFycmF5LnByb3RvdHlwZS50b0xv Y2FsZVN0cmluZyB2aXNpdHMgZWxlbWVudHMgaW4gd3Jvbmcgb3JkZXIgdW5kZXIgY2VydGFpbiBj b25kaXRpb25zCkluZGV4OiBTb3VyY2UvSmF2YVNjcmlwdENvcmUvaGVhcC9Db3BpZWRTcGFjZS5j cHAKPT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PQotLS0gU291cmNlL0phdmFTY3JpcHRDb3JlL2hlYXAvQ29waWVkU3BhY2Uu Y3BwCShyZXZpc2lvbiAxMTAzNTkpCisrKyBTb3VyY2UvSmF2YVNjcmlwdENvcmUvaGVhcC9Db3Bp ZWRTcGFjZS5jcHAJKHdvcmtpbmcgY29weSkKQEAgLTcxLDggKzcxLDggQEAgQ2hlY2tlZEJvb2xl YW4gQ29waWVkU3BhY2U6OnRyeUFsbG9jYXRlTwogewogICAgIEFTU0VSVChpc092ZXJzaXplKGJ5 dGVzKSk7CiAgICAgCi0gICAgc2l6ZV90IGJsb2NrU2l6ZSA9IFdURjo6cm91bmRVcFRvTXVsdGlw bGVPZjxzX3BhZ2VTaXplPihzaXplb2YoQ29waWVkQmxvY2spICsgYnl0ZXMpOwotICAgIFBhZ2VB bGxvY2F0aW9uQWxpZ25lZCBhbGxvY2F0aW9uID0gUGFnZUFsbG9jYXRpb25BbGlnbmVkOjphbGxv Y2F0ZShibG9ja1NpemUsIHNfcGFnZVNpemUsIE9TQWxsb2NhdG9yOjpKU0dDSGVhcFBhZ2VzKTsK KyAgICBzaXplX3QgYmxvY2tTaXplID0gV1RGOjpyb3VuZFVwVG9NdWx0aXBsZU9mKFdURjo6cGFn ZVNpemUoKSwgc2l6ZW9mKENvcGllZEJsb2NrKSArIGJ5dGVzKTsKKyAgICBQYWdlQWxsb2NhdGlv bkFsaWduZWQgYWxsb2NhdGlvbiA9IFBhZ2VBbGxvY2F0aW9uQWxpZ25lZDo6YWxsb2NhdGUoYmxv Y2tTaXplLCBXVEY6OnBhZ2VTaXplKCksIE9TQWxsb2NhdG9yOjpKU0dDSGVhcFBhZ2VzKTsKICAg ICBpZiAoIXN0YXRpY19jYXN0PGJvb2w+KGFsbG9jYXRpb24pKSB7CiAgICAgICAgICpvdXRQdHIg PSAwOwogICAgICAgICByZXR1cm4gZmFsc2U7CkluZGV4OiBTb3VyY2UvSmF2YVNjcmlwdENvcmUv aGVhcC9Db3BpZWRTcGFjZS5oCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIFNvdXJjZS9KYXZhU2NyaXB0Q29yZS9o ZWFwL0NvcGllZFNwYWNlLmgJKHJldmlzaW9uIDExMDM1OSkKKysrIFNvdXJjZS9KYXZhU2NyaXB0 Q29yZS9oZWFwL0NvcGllZFNwYWNlLmgJKHdvcmtpbmcgY29weSkKQEAgLTM1LDYgKzM1LDcgQEAK ICNpbmNsdWRlIDx3dGYvSGFzaFNldC5oPgogI2luY2x1ZGUgPHd0Zi9PU0FsbG9jYXRvci5oPgog I2luY2x1ZGUgPHd0Zi9QYWdlQWxsb2NhdGlvbkFsaWduZWQuaD4KKyNpbmNsdWRlIDx3dGYvUGFn ZUJsb2NrLmg+CiAjaW5jbHVkZSA8d3RmL1N0ZExpYkV4dHJhcy5oPgogI2luY2x1ZGUgPHd0Zi9U aHJlYWRpbmdQcmltaXRpdmVzLmg+CiAKQEAgLTExNiw4ICsxMTcsNiBAQCBwcml2YXRlOgogICAg IHNpemVfdCBtX251bWJlck9mTG9hbmVkQmxvY2tzOwogCiAgICAgc3RhdGljIGNvbnN0IHNpemVf dCBzX21heEFsbG9jYXRpb25TaXplID0gMzIgKiBLQjsKLSAgICBzdGF0aWMgY29uc3Qgc2l6ZV90 IHNfcGFnZVNpemUgPSA0ICogS0I7Ci0gICAgc3RhdGljIGNvbnN0IHNpemVfdCBzX3BhZ2VNYXNr ID0gfihzX3BhZ2VTaXplIC0gMSk7CiAgICAgc3RhdGljIGNvbnN0IHNpemVfdCBzX2luaXRpYWxC bG9ja051bSA9IDE2OwogICAgIHN0YXRpYyBjb25zdCBzaXplX3Qgc19ibG9ja01hc2sgPSB+KEhl YXBCbG9jazo6c19ibG9ja1NpemUgLSAxKTsKIH07CkluZGV4OiBTb3VyY2UvSmF2YVNjcmlwdENv cmUvaGVhcC9Db3BpZWRTcGFjZUlubGluZU1ldGhvZHMuaAo9PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBTb3VyY2Uv SmF2YVNjcmlwdENvcmUvaGVhcC9Db3BpZWRTcGFjZUlubGluZU1ldGhvZHMuaAkocmV2aXNpb24g MTEwMzU5KQorKysgU291cmNlL0phdmFTY3JpcHRDb3JlL2hlYXAvQ29waWVkU3BhY2VJbmxpbmVN ZXRob2RzLmgJKHdvcmtpbmcgY29weSkKQEAgLTE3MSw3ICsxNzEsNyBAQCBpbmxpbmUgYm9vbCBD b3BpZWRTcGFjZTo6aXNQaW5uZWQodm9pZCogCiAKIGlubGluZSBDb3BpZWRCbG9jayogQ29waWVk U3BhY2U6Om92ZXJzaXplQmxvY2tGb3Iodm9pZCogcHRyKQogewotICAgIHJldHVybiByZWludGVy cHJldF9jYXN0PENvcGllZEJsb2NrKj4ocmVpbnRlcnByZXRfY2FzdDxzaXplX3Q+KHB0cikgJiBz X3BhZ2VNYXNrKTsKKyAgICByZXR1cm4gcmVpbnRlcnByZXRfY2FzdDxDb3BpZWRCbG9jayo+KHJl aW50ZXJwcmV0X2Nhc3Q8c2l6ZV90PihwdHIpICYgV1RGOjpwYWdlTWFzaygpKTsKIH0KIAogaW5s aW5lIENvcGllZEJsb2NrKiBDb3BpZWRTcGFjZTo6YmxvY2tGb3Iodm9pZCogcHRyKQpJbmRleDog U291cmNlL0phdmFTY3JpcHRDb3JlL3d0Zi9QYWdlQmxvY2suY3BwCj09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIFNv dXJjZS9KYXZhU2NyaXB0Q29yZS93dGYvUGFnZUJsb2NrLmNwcAkocmV2aXNpb24gMTEwMzU5KQor KysgU291cmNlL0phdmFTY3JpcHRDb3JlL3d0Zi9QYWdlQmxvY2suY3BwCSh3b3JraW5nIGNvcHkp CkBAIC0zOCw2ICszOCw3IEBACiBuYW1lc3BhY2UgV1RGIHsKIAogc3RhdGljIHNpemVfdCBzX3Bh Z2VTaXplOworc3RhdGljIHNpemVfdCBzX3BhZ2VNYXNrOwogCiAjaWYgT1MoVU5JWCkKIApAQCAt NjcsNCArNjgsMTEgQEAgc2l6ZV90IHBhZ2VTaXplKCkKICAgICByZXR1cm4gc19wYWdlU2l6ZTsK IH0KIAorc2l6ZV90IHBhZ2VNYXNrKCkKK3sKKyAgICBpZiAoIXNfcGFnZU1hc2spCisgICAgICAg IHNfcGFnZU1hc2sgPSB+KHBhZ2VTaXplKCkgLSAxKTsKKyAgICByZXR1cm4gc19wYWdlTWFzazsK K30KKwogfSAvLyBuYW1lc3BhY2UgV1RGCkluZGV4OiBTb3VyY2UvSmF2YVNjcmlwdENvcmUvd3Rm L1BhZ2VCbG9jay5oCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT0KLS0tIFNvdXJjZS9KYXZhU2NyaXB0Q29yZS93dGYvUGFn ZUJsb2NrLmgJKHJldmlzaW9uIDExMDM1OSkKKysrIFNvdXJjZS9KYXZhU2NyaXB0Q29yZS93dGYv UGFnZUJsb2NrLmgJKHdvcmtpbmcgY29weSkKQEAgLTI5LDYgKzI5LDcgQEAKIG5hbWVzcGFjZSBX VEYgewogCiBXVEZfRVhQT1JUX1BSSVZBVEUgc2l6ZV90IHBhZ2VTaXplKCk7CitXVEZfRVhQT1JU X1BSSVZBVEUgc2l6ZV90IHBhZ2VNYXNrKCk7CiBpbmxpbmUgYm9vbCBpc1BhZ2VBbGlnbmVkKHZv aWQqIGFkZHJlc3MpIHsgcmV0dXJuICEocmVpbnRlcnByZXRfY2FzdDxpbnRwdHJfdD4oYWRkcmVz cykgJiAocGFnZVNpemUoKSAtIDEpKTsgfQogaW5saW5lIGJvb2wgaXNQYWdlQWxpZ25lZChzaXpl X3Qgc2l6ZSkgeyByZXR1cm4gIShzaXplICYgKHBhZ2VTaXplKCkgLSAxKSk7IH0KIGlubGluZSBi b29sIGlzUG93ZXJPZlR3byhzaXplX3Qgc2l6ZSkgeyByZXR1cm4gIShzaXplICYgKHNpemUgLSAx KSk7IH0KSW5kZXg6IFNvdXJjZS9KYXZhU2NyaXB0Q29yZS93dGYvU3RkTGliRXh0cmFzLmgKPT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PQotLS0gU291cmNlL0phdmFTY3JpcHRDb3JlL3d0Zi9TdGRMaWJFeHRyYXMuaAkocmV2 aXNpb24gMTEwMzU5KQorKysgU291cmNlL0phdmFTY3JpcHRDb3JlL3d0Zi9TdGRMaWJFeHRyYXMu aAkod29ya2luZyBjb3B5KQpAQCAtMTU0LDEyICsxNTQsMTYgQEAgdGVtcGxhdGU8dHlwZW5hbWUg VCwgc2l6ZV90IFNpemU+IGNoYXIgKAogI2RlZmluZSBXVEZfQVJSQVlfTEVOR1RIKGFycmF5KSBz aXplb2YoOjpXVEY6OkFycmF5TGVuZ3RoSGVscGVyRnVuY3Rpb24oYXJyYXkpKQogCiAvLyBFZmZp Y2llbnQgaW1wbGVtZW50YXRpb24gdGhhdCB0YWtlcyBhZHZhbnRhZ2Ugb2YgcG93ZXJzIG9mIHR3 by4KK2lubGluZSBzaXplX3Qgcm91bmRVcFRvTXVsdGlwbGVPZihzaXplX3QgZGl2aXNvciwgc2l6 ZV90IHgpCit7CisgICAgQVNTRVJUKGRpdmlzb3IgJiYgIShkaXZpc29yICYgKGRpdmlzb3IgLSAx KSkpOworICAgIHNpemVfdCByZW1haW5kZXJNYXNrID0gZGl2aXNvciAtIDE7IFwKKyAgICByZXR1 cm4gKHggKyByZW1haW5kZXJNYXNrKSAmIH5yZW1haW5kZXJNYXNrOworfQogdGVtcGxhdGU8c2l6 ZV90IGRpdmlzb3I+IGlubGluZSBzaXplX3Qgcm91bmRVcFRvTXVsdGlwbGVPZihzaXplX3QgeCkK IHsKICAgICBDT01QSUxFX0FTU0VSVChkaXZpc29yICYmICEoZGl2aXNvciAmIChkaXZpc29yIC0g MSkpLCBkaXZpc29yX2lzX2FfcG93ZXJfb2ZfdHdvKTsKLQotICAgIHNpemVfdCByZW1haW5kZXJN YXNrID0gZGl2aXNvciAtIDE7Ci0gICAgcmV0dXJuICh4ICsgcmVtYWluZGVyTWFzaykgJiB+cmVt YWluZGVyTWFzazsKKyAgICByZXR1cm4gcm91bmRVcFRvTXVsdGlwbGVPZihkaXZpc29yLCB4KTsK IH0KIAogZW51bSBCaW5hcnlTZWFyY2hNb2RlIHsK