r252114 - [modules] If we're given a module file, via -fmodule-file=, for a module, but

[Ben Langmuir via cfe-commits]

FYI I changed this to early-exit again in r260563 to fix implicit builds.  It’s not safe to keep reading if we haven’t told ReadOptionsBlock to expect failures, since it will itself early-exit, leaving the stream in the middle of a block.  It’s also a performance advantage to early exit when doing implicit builds.

Ben

> On Jan 8, 2016, at 2:46 PM, Adrian Prantl via cfe-commits <cfe-commits at lists.llvm.org> wrote:
> 
>> 
>> On Jan 8, 2016, at 2:39 PM, Adrian Prantl <aprantl at apple.com> wrote:
>> 
>> Hi Richard,
>> 
>> This change 
>>> @@ -2239,16 +2240,21 @@ ASTReader::ReadControlBlock(ModuleFile &F,
>> [...]
>>> -          if (!DisableValidation && Result != Success &&
>>> -              (Result != ConfigurationMismatch || !AllowConfigurationMismatch))
>>> +          if (DisableValidation ||
>>> +              (AllowConfigurationMismatch && Result == ConfigurationMismatch))
>>> +            Result = Success;
>>> +
>>> +          // If we've diagnosed a problem, we're done.
>>> +          if (Result != Success &&
>>> +              isDiagnosedResult(Result, ClientLoadCapabilities))
>>>            return Result;
>> 
>> either causes or uncovers a bug:
>> 
>>> CC=/Volumes/Data/llvm/_build.ninja.debug/bin/clang # r256948
>>> rm -rf cache && mkdir cache
>>> rm -rf Test && mkdir Test
>>> echo 'module Test {
>>> umbrella header "Test.h"
>>> }' >Test/module.modulemap
>>> touch Test/Test.h
>>> echo '#import <Test/Test.h>'>2.m
>>> 
>>> clang -x objective-c -fmodules -fmodules-cache-path=cache  -DA=0 -I. -c 2.m -o 1.o
>>> clang -x objective-c -fmodules -fmodules-cache-path=cache -Werror -DA=0 -I. -c 2.m -o 2.o
>>> 
>> 
>> 
>> After encountering a configuration mismatch or out-of-date error, we now continue instead of returning early and subsequently crash in
>> 
>> ASTReader::ReadControlBlock()
>>   ASTReader::getInputFile()
>>     Cursor.JumpToBit(F.InputFileOffsets[ID-1]);
>> 
>> I’ll keep digging deeper, but I thought you may have an intuition of what’s going on here.
>> Is the behavior change intentional? From the commit message it sounds as if implicit module builds shouldn’t be affected.
> 
> In my particular crash isDiagnosedResult returns false for an OutOfDate result thus skipping the early return.
> 
> -- adrian
> _______________________________________________
> cfe-commits mailing list
> cfe-commits at lists.llvm.org <mailto:cfe-commits at lists.llvm.org>
> http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits <http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20160211/78b44278/attachment.html>