similar to: [LLVMdev] All CallInsts mayHaveSideEffects

Does any real code benefit from dead code eliminating read-only functions? Tom ----- Original Message ----- From: "Reid Kleckner" <rnk at mit.edu> To: "Thomas B. Jablin" <tjablin at CS.Princeton.EDU> Cc: "LLVM Developers Mailing List" <llvmdev at cs.uiuc.edu> Sent: Monday, May 10, 2010 9:38:47 PM GMT -05:00 US/Canada Eastern Subject: Re: [LLVMdev]

Hi Nadav, > I almost missed your email because you replied only to the list. I understand your argument. I think that my fix addresses a part of it. It says that instructions that do not return should not be removed. > The current implementation of mayReturn is to check the 'noreturn' function attribute. Are you suggesting to add a new attribute that is called 'mustreturn'

Currently LLVM removes infinite recursion in readnone functions as dead code. I assume this falls under the same language rules that allow removal of infinite loops. But I need to be certain. Can someone confirm? Standard citations welcome. Otherwise this is wrong: bool mayHaveSideEffects() const { return mayWriteToMemory() || mayThrow() || !mayReturn(); } Note: For non-C support, it

LLVM today does not clearly specify if a function specified to not write to memory (i.e. readonly or readnone) is allowed to throw exceptions. LangRef is ambiguous on this issue. The normative statement is "[readnone/readonly functions] cannot unwind exceptions by calling the C++ exception throwing methods" which does not decide an answer for non C++ languages. It used to say (h/t

On 19.03.2016 16:25, Mehdi Amini wrote: > Hi, > > Can you elaborate what is the impact at the IR level? > If the point is just about how you lower for you target, why are you needing an IR level attribute? You backend if free to specialize the lowering for any intrinsic regardless the IR level attributes. As I explained in my reply to Philip, what I really need is a way to get

> On Mar 21, 2016, at 9:14 PM, Mehdi Amini via llvm-dev <llvm-dev at lists.llvm.org> wrote: > > >> On Mar 21, 2016, at 8:58 AM, Nicolai Hähnle <nhaehnle at gmail.com> wrote: >> >> On 19.03.2016 16:25, Mehdi Amini wrote: >>> Hi, >>> >>> Can you elaborate what is the impact at the IR level? >>> If the point is just about

On Aug 21, 2009, at 8:46 PM, Eli Friedman wrote: > On Fri, Aug 21, 2009 at 3:29 PM, Jakub Staszak<kuba at gcc.gnu.org> > wrote: >> >> On Aug 21, 2009, at 7:31 PM, Eli Friedman wrote: >> >>> On Fri, Aug 21, 2009 at 2:06 PM, Jakub Staszak<kuba at gcc.gnu.org> >>> wrote: >>>> >>>> Hello, >>>>

On Aug 21, 2009, at 10:02 PM, Eli Friedman wrote: > On Fri, Aug 21, 2009 at 4:53 PM, Jakub Staszak<kuba at gcc.gnu.org> > wrote: >> >> On Aug 21, 2009, at 8:46 PM, Eli Friedman wrote: >> >>> On Fri, Aug 21, 2009 at 3:29 PM, Jakub Staszak<kuba at gcc.gnu.org> >>> wrote: >>>> >>>> On Aug 21, 2009, at 7:31 PM, Eli

On Aug 21, 2009, at 10:27 PM, Eli Friedman wrote: > On Fri, Aug 21, 2009 at 5:05 PM, Jakub Staszak<kuba at gcc.gnu.org> > wrote: >> >> On Aug 21, 2009, at 10:02 PM, Eli Friedman wrote: >> >>> On Fri, Aug 21, 2009 at 4:53 PM, Jakub Staszak<kuba at gcc.gnu.org> >>> wrote: >>>> >>>> On Aug 21, 2009, at 8:46 PM, Eli

On Fri, Aug 21, 2009 at 4:53 PM, Jakub Staszak<kuba at gcc.gnu.org> wrote: > > On Aug 21, 2009, at 8:46 PM, Eli Friedman wrote: > >> On Fri, Aug 21, 2009 at 3:29 PM, Jakub Staszak<kuba at gcc.gnu.org> wrote: >>> >>> On Aug 21, 2009, at 7:31 PM, Eli Friedman wrote: >>> >>>> On Fri, Aug 21, 2009 at 2:06 PM, Jakub Staszak<kuba at

On Fri, Aug 21, 2009 at 5:05 PM, Jakub Staszak<kuba at gcc.gnu.org> wrote: > > On Aug 21, 2009, at 10:02 PM, Eli Friedman wrote: > >> On Fri, Aug 21, 2009 at 4:53 PM, Jakub Staszak<kuba at gcc.gnu.org> wrote: >>> >>> On Aug 21, 2009, at 8:46 PM, Eli Friedman wrote: >>> >>>> On Fri, Aug 21, 2009 at 3:29 PM, Jakub Staszak<kuba at

On Fri, Aug 21, 2009 at 5:47 PM, Jakub Staszak<kuba at gcc.gnu.org> wrote: > > On Aug 21, 2009, at 10:27 PM, Eli Friedman wrote: > >> On Fri, Aug 21, 2009 at 5:05 PM, Jakub Staszak<kuba at gcc.gnu.org> wrote: >>> >>> On Aug 21, 2009, at 10:02 PM, Eli Friedman wrote: >>> >>>> On Fri, Aug 21, 2009 at 4:53 PM, Jakub Staszak<kuba at

While improving ADCE, i notice that for declare i32 @strlen(i8*) readnone define i32 @test() { ; invoke of pure function should not be deleted! invoke i32 @strlen( i8* null ) readnone to label %Cont unwind label %Other ; <i32>:1 [#uses=0] Cont: ; preds = %0 ret i32 0 Other: ; preds = %0 %exn = landingpad {i8*, i32} personality i32 (...)* @__gxx_personality_v0

Hello, Currently, llvm::isSafeToSpeculativelyExecute() always returns false for Call instructions. This has actual performance implications, because loop-invariant code motion makes this check, and will never hoist instructions that are not safe to speculatively execute. Unfortunately, there is currently no way to signal to LICM that a function is safe to speculatively execute. The

Hello, This patch fixes PR4174. Two test-cases included: original one from bugzilla and a little bit complicated made be myself. It seems that LoopIndexSplit doesn't handle some cases, I'll try to send some patch this week. Regards -- Jakub Staszak -------------- next part -------------- A non-text attachment was scrubbed... Name: pr4174.patch Type: application/octet-stream Size:

for (BasicBlock::reverse_iterator I = BB.rbegin(), E = BB.rend(); I != E; ) { Instruction& inst = *I; ++I; <-- iterator should be advanced to the previous instruction // Happens to be an Instruction::SExt. // Works fine if I iterate forwards if (isInstructionTriviallyDead(&inst, TLI)) inst.eraseFromParent(); } Sorry for the inexperienced question, but

On Aug 21, 2009, at 7:31 PM, Eli Friedman wrote: > On Fri, Aug 21, 2009 at 2:06 PM, Jakub Staszak<kuba at gcc.gnu.org> > wrote: >> Hello, >> >> This patch fixes PR4174. Two test-cases included: original one from >> bugzilla >> and a little bit complicated made be myself. > > I think you want isSafeToSpeculativelyExecute rather than >

Hi I am currently trying to schedule instructions with my own algorithm. For that i need to get the data dependency between the instructions. So currently i am dooing s.t. like: for(BasicBlock::iterator j=B.begin(),bbe=B.end();j!=bbe;++j) { InstructionList.push_back(j); if (const AllocaInst *AI = dyn_cast<AllocaInst>(j)) {

On Fri, 12 May 2006, Silken Tiger wrote: > To find the first instructions which are not depending on others results. So > far it seems to be working but i am missing instructions like: > %tmp.1 = seteq int %argc, 2 ; <bool> [#uses=1] > There seems only an function like llvm::Instruction::mayWriteToMemory > but nothing like llvm::Instruction::mayReadFromMemory or

The Instruction::mayWriteToMemory() member function is quite useful (I'm currently writing code that will break up basic blocks at read/write instructions). Is there a safe way to do the same thing for reads from memory? It's easy enough to check for the relevant instructions, but I thought this might be a nice-to-have in a future version because it separates code like mine from