[llvm] r222971 - Revert r222957 "Replace std::map<K, V*> with std::map<K, V> to handle ownership and deletion of the values."
Craig Topper
craig.topper at gmail.com
Sat Nov 29 17:28:34 PST 2014
Ok one of the bots caught it with clang, but my local build passed. Weird.
On Sat, Nov 29, 2014 at 7:20 PM, Craig Topper <craig.topper at gmail.com>
wrote:
> Author: ctopper
> Date: Sat Nov 29 19:20:17 2014
> New Revision: 222971
>
> URL: http://llvm.org/viewvc/llvm-project?rev=222971&view=rev
> Log:
> Revert r222957 "Replace std::map<K, V*> with std::map<K, V> to handle
> ownership and deletion of the values."
>
> Upon further review I think the MultiClass is being copied into the map
> instead of being moved due to the copy constructor on the nested Record
> type. This ultimately got exposed when the vector in DefPrototype vector
> was changed to hold unique_ptrs in another commit. This caused gcc 4.7 to
> fail due to the use of the copy constructor on unique_ptr with the error
> pointing back to one of the insert calls from this commit. Not sure why
> clang was able to build.
>
> This reverts commit 710cdf729f84b428bf41aa8d32dbdb35fff79fde.
>
> Modified:
> llvm/trunk/lib/TableGen/TGParser.cpp
> llvm/trunk/lib/TableGen/TGParser.h
>
> Modified: llvm/trunk/lib/TableGen/TGParser.cpp
> URL:
> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/TableGen/TGParser.cpp?rev=222971&r1=222970&r2=222971&view=diff
>
> ==============================================================================
> --- llvm/trunk/lib/TableGen/TGParser.cpp (original)
> +++ llvm/trunk/lib/TableGen/TGParser.cpp Sat Nov 29 19:20:17 2014
> @@ -459,12 +459,12 @@ MultiClass *TGParser::ParseMultiClassID(
> return nullptr;
> }
>
> - auto it = MultiClasses.find(Lex.getCurStrVal());
> - if (it == MultiClasses.end())
> + MultiClass *Result = MultiClasses[Lex.getCurStrVal()];
> + if (!Result)
> TokError("Couldn't find multiclass '" + Lex.getCurStrVal() + "'");
>
> Lex.Lex();
> - return &it->second;
> + return Result;
> }
>
> /// ParseSubClassReference - Parse a reference to a subclass or to a
> templated
> @@ -2290,13 +2290,11 @@ bool TGParser::ParseMultiClass() {
> return TokError("expected identifier after multiclass for name");
> std::string Name = Lex.getCurStrVal();
>
> - auto Result =
> - MultiClasses.insert(std::make_pair(Name,
> - MultiClass(Name,
> Lex.getLoc(),Records)));
> - if (!Result.second)
> + if (MultiClasses.count(Name))
> return TokError("multiclass '" + Name + "' already defined");
> - CurMultiClass = &Result.first->second;
>
> + CurMultiClass = MultiClasses[Name] = new MultiClass(Name,
> + Lex.getLoc(),
> Records);
> Lex.Lex(); // Eat the identifier.
>
> // If there are template args, parse them.
> @@ -2557,9 +2555,8 @@ bool TGParser::ParseDefm(MultiClass *Cur
> // To instantiate a multiclass, we need to first get the multiclass,
> then
> // instantiate each def contained in the multiclass with the
> SubClassRef
> // template parameters.
> - auto it = MultiClasses.find(Ref.Rec->getName());
> - assert(it != MultiClasses.end() && "Didn't lookup multiclass
> correctly?");
> - MultiClass *MC = &it->second;
> + MultiClass *MC = MultiClasses[Ref.Rec->getName()];
> + assert(MC && "Didn't lookup multiclass correctly?");
> std::vector<Init*> &TemplateVals = Ref.TemplateArgs;
>
> // Verify that the correct number of template arguments were
> specified.
>
> Modified: llvm/trunk/lib/TableGen/TGParser.h
> URL:
> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/TableGen/TGParser.h?rev=222971&r1=222970&r2=222971&view=diff
>
> ==============================================================================
> --- llvm/trunk/lib/TableGen/TGParser.h (original)
> +++ llvm/trunk/lib/TableGen/TGParser.h Sat Nov 29 19:20:17 2014
> @@ -55,7 +55,7 @@ namespace llvm {
> class TGParser {
> TGLexer Lex;
> std::vector<std::vector<LetRecord> > LetStack;
> - std::map<std::string, MultiClass> MultiClasses;
> + std::map<std::string, MultiClass*> MultiClasses;
>
> /// Loops - Keep track of any foreach loops we are within.
> ///
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>
--
~Craig
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20141129/48b9dc67/attachment.html>
More information about the llvm-commits
mailing list