[LLVMdev] LLVM: Very simple question

[James Molloy]

Hi,

Have you tried using the "Show LLVM C++ API code" option on the online demo?

I could explain what to do to you, but you might understand it better with the interactive example. For your code snippet, it spits out

... snip ...
    AllocaInst* ptr_retval = new AllocaInst(IntegerType::get(mod->getContext(), 32), "retval", label_entry);
    ptr_retval->setAlignment(4);
    AllocaInst* ptr_var = new AllocaInst(ArrayTy_3, "var", label_entry);
    ptr_var->setAlignment(1);
    AllocaInst* ptr_offset = new AllocaInst(IntegerType::get(mod->getContext(), 32), "offset", label_entry);
    ptr_offset->setAlignment(4);
    AllocaInst* ptr_val = new AllocaInst(IntegerType::get(mod->getContext(), 32), "val", label_entry);
    ptr_val->setAlignment(4);
     new StoreInst(const_int32_5, ptr_offset, false, label_entry);
    LoadInst* int32_tmp = new LoadInst(ptr_offset, "tmp", false, label_entry);
    CastInst* int64_idx_ext = new ZExtInst(int32_tmp, IntegerType::get(mod->getContext(), 64), "idx.ext", label_entry);
    GetElementPtrInst* ptr_add_ptr = GetElementPtrInst::Create(ptr_var, int64_idx_ext, "add.ptr", label_entry);
    CastInst* ptr_8 = new BitCastInst(ptr_add_ptr, PointerTy_1, "", label_entry);
    LoadInst* int32_tmp1 = new LoadInst(ptr_8, "tmp1", false, label_entry);
     new StoreInst(int32_tmp1, ptr_val, false, label_entry);
    LoadInst* int32_tmp2 = new LoadInst(ptr_offset, "tmp2", false, label_entry);
    CastInst* int64_idx_ext3 = new ZExtInst(int32_tmp2, IntegerType::get(mod->getContext(), 64), "idx.ext3", label_entry);
    GetElementPtrInst* ptr_add_ptr4 = GetElementPtrInst::Create(ptr_var, int64_idx_ext3, "add.ptr4", label_entry);
    CastInst* ptr_10 = new BitCastInst(ptr_add_ptr4, PointerTy_1, "", label_entry);
     new StoreInst(const_int32_6, ptr_10, false, label_entry);
    LoadInst* int32_12 = new LoadInst(ptr_retval, "", false, label_entry);
    ReturnInst::Create(mod->getContext(), int32_12, label_entry);
... snip ...

With the important bits being the AllocaInst, StoreInsts, LoadInsts and BitCasts. With simpler code you'd get more obvious output.

You can ignore the ->setAlignment()'s and just use the native alignment.

If this doesn't help any then I'd be happy to explain more.

Cheers,

James

From: llvmdev-bounces at cs.uiuc.edu [mailto:llvmdev-bounces at cs.uiuc.edu] On Behalf Of Semion Prihodko
Sent: 19 August 2011 16:01
To: llvmdev at cs.uiuc.edu
Subject: [LLVMdev] LLVM: Very simple question

Hi, guys. I'm a newbie to LLVM and have a very simple question.

Which instructions should I use (in terms of IRBuilder calls) to allocate an array of bytes in stack (alloca?), then to work with it (from a given offset) as with integer (bitcast?). I mean something like that:

unsigned char var[8];
unsigned int offset = 3;
int val = *(int*)(&var+offset); /* read */
*(int*)(&var+offset) = 12; /* write */

Yes, I can convert the code above to assembly language using LLVM online demo, but how can I perform that programmatically using IRBuilder?

Are there some LLVM code examples less trivial than Kaleidoscope?

Thanks in advance.

-- IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20110819/1b744376/attachment.html>