[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-commits
Subject:    Re: [kdelibs/KDE/4.8] kjs: KJS: fix binary operator use instead of logical one
From:       "Rolf Eike Beer" <kde () opensource ! sf-tec ! de>
Date:       2012-05-16 6:33:27
Message-ID: ef86ec192be9d3fc3eb54b03dfe5df9d.squirrel () webmail ! sf-mail ! de
[Download RAW message or body]

> On 5/15/12, Rolf Eike Beer <kde@opensource.sf-tec.de> wrote:
>> Git commit 2efcba3c713546c6dd3859b527611a52db2f5538 by Rolf Eike Beer.
>> Committed on 15/05/2012 at 20:19.
>> Pushed by dakon into branch 'KDE/4.8'.
>
> Unless you have benchmarks showing this is not a performance regression,
> please revert.

Sorry, that has nothing to do with performance, it's just deobfuscation. I
compared the assembler output of both versions. For every optimization
beginning with O1 the output is exactly the same.

>> KJS: fix binary operator use instead of logical one
>>
>> M  +1    -1    kjs/collector.cpp
>>
>> http://commits.kde.org/kdelibs/2efcba3c713546c6dd3859b527611a52db2f5538
>>
>> diff --git a/kjs/collector.cpp b/kjs/collector.cpp
>> index 7a7190c..70e4757 100644
>> --- a/kjs/collector.cpp
>> +++ b/kjs/collector.cpp
>> @@ -467,7 +467,7 @@ void Collector::markStackObjectsConservatively(void
>> *start, void *end)
>>        uintptr_t offset = reinterpret_cast<uintptr_t>(x) &
>> BLOCK_OFFSET_MASK;
>>        CollectorBlock* blockAddr = reinterpret_cast<CollectorBlock*>(x -
>> offset);
>>        for (size_t block = 0; block < usedBlocks; block++) {
>> -        if ((blocks[block] == blockAddr) & (offset <= lastCellOffset))
>> {
>> +        if ((blocks[block] == blockAddr) && (offset <= lastCellOffset))
>> {
>>            if (((CollectorCell *)x)->u.freeCell.zeroIfFree != 0) {
>>              JSCell *imp = reinterpret_cast<JSCell *>(x);
>>              if (!imp->marked())
>>
>

[prev in list] [next in list] [prev in thread] [next in thread]