similar to: [LLVMdev] Which is more compact, .bc or .ll.gz? And what might be even more compact?

On Tue, 14 Dec 2010 10:55:09 -0000 "Edmund Grimley-Evans" <Edmund.Grimley-Evans at arm.com> wrote: > According to the few tests I did, .ll.gz is more compact: > > 1.00 LLVM bitcode (.bc) > 0.80 Gzipped LLVM bitcode (.bc.gz) > 4.13 LLVM assembly (.ll) > 0.68 Gzipped LLVM assembly (.ll.gz) > > However, there's not much in it, considering that a

I just got this error message from the GNU assembler: Error: alignment too large: 15 assumed Which made me laugh at first. The corresponding input line was: .align 16 Apparently what's going on here is that ".align 16" is ambiguous: on some architectures it means ".balign 16", and on some it means ".p2align 16", which would mean ".balign 65536" if

I'm getting a lot of this from make check: g++: unrecognized option `-Qunused-arguments' This presumably comes from commit 94666 (test/LLVMC/ExternOptions.td and other files). That option is specific to clang, isn't it? Should %compile_cxx be picking up clang instead of my system's g++? -- IMPORTANT NOTICE: The contents of this email and any attachments are confidential and

I think this test case demonstrates it: ; RUN: llc -march=thumb -mcpu=cortex-a8 -mtriple=thumbv7-eabi -float-abi=hard < %s | FileCheck %s define <4 x i64> @f_4_i64(<4 x i64> %a, <4 x i64> %b) nounwind { ; CHECK: vadd.i64 %y = add <4 x i64> %a, %b ret <4 x i64> %y } (I hope I got that right.) -- IMPORTANT NOTICE: The contents of this email and any

The log message for revision 122213 says: > Change the X86 backend to stop using the evil ADDC/ADDE/SUBC/SUBE nodes (which > their carry depenedencies with MVT::Flag operands) and use clean and beautiful > EFLAGS dependences instead. (MVT::Flag has since been renamed to MVT::Glue.) That revision made bug 8404 go away. Am I right in thinking that one of the problems with MVT::Glue is

I don't think this has ever worked. Evan On May 13, 2010, at 10:03 AM, Edmund Grimley-Evans wrote: > I think this test case demonstrates it: > > ; RUN: llc -march=thumb -mcpu=cortex-a8 -mtriple=thumbv7-eabi -float-abi=hard < %s | FileCheck %s > > define <4 x i64> @f_4_i64(<4 x i64> %a, <4 x i64> %b) nounwind { > ; CHECK: vadd.i64 > %y = add <4

Could some kind soul point me at the documentation, or the appropriate header file, or an example of the best way to write a pass that modifies the IR? Suppose, for example, that I wanted to replace every call to the intrinsic i1 with a call to i2 followed by a call to i3: i1(X) -> i3(i2(X)) I'm currently playing around with a class that inherits from FunctionPass and InstVisitor and I

It used to. I just checked with r90370. deep On Thu, May 13, 2010 at 11:53 PM, Evan Cheng <evan.cheng at apple.com> wrote: > I don't think this has ever worked. > > Evan > > On May 13, 2010, at 10:03 AM, Edmund Grimley-Evans wrote: > >> I think this test case demonstrates it: >> >> ; RUN: llc -march=thumb -mcpu=cortex-a8 -mtriple=thumbv7-eabi

> Having a look at clang's source, you can find in > "lib/CodeGen/BackendUtil.cpp" the functions where clang builds the > passes to emit code. The optimization passes used are there and you can > simulate them via the "opt" utility, by running each pass one at a time. > LLVM also declares standard module/function passes on >

Here's a patch to fix some bugs in my code, plus a header file I probably forget to send in before. * Missing header file charset.h. It's not needed outside vorbis-tools/share, so I think it can live there. * There was a minor bug in the UTF-8 decoder utf8_mbtowc(). * I've added some more tests in charset_test.c in an attempt to convince myself that there aren't any more bugs.

With a modified copy of LLVM (so it's probably my fault) I'm getting an assertion failure because isTwoAddrUse (in TwoAddressInstructionPass.cpp) is being called with a MachineInstr MI such that MI.getNumOperands() is 2, but MI.getDesc().getNumOperands() is 5. The assertion fails when that function calls MI.getOperand(2). My question is: is isTwoAddrUse doing the right thing here? static

Using GCC 3.4.6 20060404 (Red Hat 3.4.6-11) I've had this problem with the ALWAYS_INLINE directive in SelectionDAGISel.cpp: /arm/scratch/egrimley/llvm.svn/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp: In function `bool CheckChildType(const unsigned char*, unsigned int&, llvm::SDValue, const llvm::TargetLowering&, unsigned int)':

My experimental code calls DAG.getNode to construct a unary node with a flag result. Unfortunately the argument turns out to be constant, so lib/CodeGen/SelectionDAG/SelectionDAG.cpp:2332 calls VT.getSizeInBits on the flag type, which isSimple(), so we call V.getSizeInBits at ValueTypes.h:560 and fail at ValueTypes.h:240: clang: .../include/llvm/CodeGen/ValueTypes.h:240: unsigned int

When I'm trying to understand how or why the back end does something I've found the -view-* options to llc to be useful. It's not ideal, because "clang" doesn't take those options, so I have to invoke first "clang" and then "llc", and sometimes that combination gives different final code from invoking just "clang", but usually I can construct

Assuming you're working with an ARM target, you may also hit a problem with the alignment option on the .comm directive. Attached is a first-cut patch for this latter problem. deep On Mon, Jan 25, 2010 at 5:42 PM, Edmund Grimley Evans <Edmund.Grimley-Evans at arm.com> wrote: > I just got this error message from the GNU assembler: > > Error: alignment too large: 15 assumed

Here's a propsed heavy-duty solution for your UTF-8 problems. I'm including a patch in this message, but I'll put the new files on my web site at http://rano.org/tmp/xiph_files.tar.gz I've tested this by running vorbiscomment with and without -DHAVE_ICONV=1 in vorbis-tools/share/Makefile. It seems to work. Changed files: acinclude.m4: Add a test for nl_langinfo(CODESET). This

Hi, I'm having difficulty configuring Printing with Samba 2.2.4 and Windows 2000 clients. The samba installations is running as a PDC against LDAP. All the sharing of network file systems seem to be working perfectly.. I have a share [printers] path = /usr/spool/samba guest ok = yes printable = yes browsable = no guest ok = yes writable =

James, many thanks. Is there any linker available for Macs that has a freely available binary version? I thought maybe that GNU’s linker might fit the bill? I cannot use Apple’s linker from /usr/bin/ as it is not allowed to make external calls from a sandboxed app. Hence my interest in the LLVM lld. > On 7 May 2020, at 19:21, James Y Knight <jyknight at google.com> wrote: > > On

Hi, I've been thinking about how to keep the line number with the llvm transform/Analysis passes. Basically, I agree with Chris's notes ( http://www.nondot.org/sabre/LLVMNotes/DebugInfoImprovements.txt), and I will follow his way to turn on the line number information when optimization enabled. Here is a detailed proposal: 1. Introduction At the time of this writing, LLVM's

I have been searching all over for the answer on all sources online and have come to the conclusion that it must be rudimentary or I am asking the wrong question. I cannot figure out how to configure the box to set the "from" address to a correct domain, as my outgoing isp will not pass mail from root@asterisk1.local, as I expect it wouldn't. Any help is appreciated, even just