Pedro Artigas
2012-Nov-16 21:26 UTC
[LLVMdev] AddressSanitizer depends on order of doFinalization
Hello All, There is an issue in AddressSanitizer today because it depends on it's doFinalization method be called before the Printer method to output the final code (that is the printer runOnModule call). This assumption seems invalid as the doFinalization method should AFAIK only be used for clean up tasks independent of activities of other passes. When I try to call doFinalization inside the MPPassManager::runOnModule method (to avoid changing the API to add doInitialization and doFinalization to the PassManager interface, as recommended by Chris) the assumption breaks as all the doFinalization methods will now be called after all the runOnModule or runOnFunction invocations. Any comments/suggestions? Thanks Pedro Artigas
Alexey Samsonov
2012-Nov-23 07:12 UTC
[LLVMdev] AddressSanitizer depends on order of doFinalization
Hi Pedro, Looks like r168484 is the first step to remove ASan instrumentation from doFinalization(). On Sat, Nov 17, 2012 at 1:26 AM, Pedro Artigas <partigas at apple.com> wrote:> Hello All, > > There is an issue in AddressSanitizer today because it depends on it's > doFinalization method be called before the Printer method to output the > final code (that is the printer runOnModule call). This assumption seems > invalid as the doFinalization method should AFAIK only be used for clean up > tasks independent of activities of other passes. > > When I try to call doFinalization inside the MPPassManager::runOnModule > method (to avoid changing the API to add doInitialization and > doFinalization to the PassManager interface, as recommended by Chris) the > assumption breaks as all the doFinalization methods will now be called > after all the runOnModule or runOnFunction invocations. > > Any comments/suggestions? > > Thanks > > Pedro Artigas > _______________________________________________ > LLVM Developers mailing list > LLVMdev at cs.uiuc.edu http://llvm.cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev >-- Alexey Samsonov, MSK -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20121123/a5c73de2/attachment.html>
Maybe Matching Threads
- [LLVMdev] GCMetadataPrinter::finishAssembly not executed?
- [LLVMdev] GCMetadataPrinter::finishAssembly not executed?
- [LLVMdev] GCMetadataPrinter::finishAssembly not executed?
- [LLVMdev] GCMetadataPrinter::finishAssembly not executed?
- [LLVMdev] GCMetadataPrinter::finishAssembly not executed?