Page MenuHomePhabricator

tinti (Vinicius Tinti)
User

Projects

User does not belong to any projects.

User Details

User Since
Jul 4 2013, 3:01 PM (380 w, 4 d)

Recent Activity

Fri, Oct 16

tinti added a comment to D85024: [llvm-objdump] Implement --prefix option.

Thanks @MaskRay and @thakis.

Fri, Oct 16, 1:53 PM · Restricted Project

Tue, Oct 13

tinti added a comment to D85024: [llvm-objdump] Implement --prefix option.

@MaskRay ping?

Tue, Oct 13, 2:51 PM · Restricted Project

Fri, Oct 9

tinti updated the diff for D85024: [llvm-objdump] Implement --prefix option.

Concise use of = in tests.
Merge ifs.
Reorder alphabetically the options.

Fri, Oct 9, 2:45 AM · Restricted Project

Thu, Oct 8

tinti updated the diff for D85024: [llvm-objdump] Implement --prefix option.

Reword.
Use James suggestions.

Thu, Oct 8, 6:35 AM · Restricted Project
tinti updated the diff for D85024: [llvm-objdump] Implement --prefix option.

Reword comments.
Substitute CHECK-TRAILING.
Remove extra sed pipe.
Add test for relative paths.

Thu, Oct 8, 5:43 AM · Restricted Project
tinti added inline comments to D85024: [llvm-objdump] Implement --prefix option.
Thu, Oct 8, 5:36 AM · Restricted Project

Wed, Oct 7

tinti updated the diff for D85024: [llvm-objdump] Implement --prefix option.

Add specific tests.
Reword.

Wed, Oct 7, 4:29 PM · Restricted Project
tinti added a comment to D88502: [llvm][Support] Delete unused is_absolute_gnu helper.

Hi @tinti this code seems to be recently introduced without any usage, and as pointed out by @certl it doesn't consider multi-byte drive letters.

Wed, Oct 7, 4:26 PM · Restricted Project

Mon, Oct 5

tinti added a comment to D85024: [llvm-objdump] Implement --prefix option.

I believe I should create more tests addressing:

Mon, Oct 5, 7:49 PM · Restricted Project

Thu, Sep 24

tinti added a comment to D87667: [Support/Path] Add path::is_absolute_gnu.

This patch added an issue in some builds:

Thu, Sep 24, 3:40 AM · Restricted Project

Wed, Sep 23

tinti added inline comments to D87667: [Support/Path] Add path::is_absolute_gnu.
Wed, Sep 23, 4:26 AM · Restricted Project
tinti updated the summary of D87667: [Support/Path] Add path::is_absolute_gnu.
Wed, Sep 23, 4:25 AM · Restricted Project
tinti updated the diff for D87667: [Support/Path] Add path::is_absolute_gnu.

Add reference to cppreference.com.
Improve comments.

Wed, Sep 23, 4:25 AM · Restricted Project

Mon, Sep 21

tinti updated the diff for D87667: [Support/Path] Add path::is_absolute_gnu.

Remove SmallVector.
Move comment to is_absolute.
Reword documentation.
Add references.

Mon, Sep 21, 1:39 PM · Restricted Project
tinti added a comment to D87667: [Support/Path] Add path::is_absolute_gnu.

I have done several rewords.

Mon, Sep 21, 1:31 PM · Restricted Project
tinti updated the diff for D87667: [Support/Path] Add path::is_absolute_gnu.
  • Drop the "_style" in the name of the function.
  • Add ":" test.
  • Add comment about LLVM is_absolute.
  • Add Vector constructor.
Mon, Sep 21, 4:14 AM · Restricted Project

Sep 18 2020

tinti updated the diff for D87667: [Support/Path] Add path::is_absolute_gnu.

Add more tests;
Simplify is_gnu_style_absolute;
Reword comments;
Reword commit message.

Sep 18 2020, 3:24 AM · Restricted Project

Sep 16 2020

tinti added inline comments to D87667: [Support/Path] Add path::is_absolute_gnu.
Sep 16 2020, 2:30 AM · Restricted Project

Sep 15 2020

tinti added a comment to D87667: [Support/Path] Add path::is_absolute_gnu.

On Windows, paths starting with drive letter pattern are absolute.

This isn't a difference is it? That's also the behaviour for is_absolute, I believe?

Sep 15 2020, 7:14 PM · Restricted Project
tinti added a comment to D87667: [Support/Path] Add path::is_absolute_gnu.
Sep 15 2020, 6:52 PM · Restricted Project

Sep 14 2020

tinti requested review of D87667: [Support/Path] Add path::is_absolute_gnu.
Sep 14 2020, 7:25 PM · Restricted Project

Sep 1 2020

tinti added a comment to D85024: [llvm-objdump] Implement --prefix option.

Should these crash tests take into account the Style?

Sep 1 2020, 5:25 PM · Restricted Project

Aug 26 2020

tinti added a comment to D85024: [llvm-objdump] Implement --prefix option.

Having given it some thought, I think the right thing to do might be to put a method alongside the existing is_absolute method called something like either is_rooted, or is_gnu_style_absolute or similar. That function would then look something like this:

bool is_rooted(StringRef Path, Style St) {
  if (is_absolute(Path, St))
    return true;
  if (real_style(St) == windows)
    return !Path.empty() && is_separator(Path.front(), St);
  return false;
}

I would probably recommend that that be added in a separate patch, with dedicated gtest unit tests and reviewers taken from among those who have recently worked in this area (plus myself). You would then rebase this patch on top of that one.

Aug 26 2020, 4:31 PM · Restricted Project

Aug 25 2020

tinti updated the diff for D85024: [llvm-objdump] Implement --prefix option.

Work done so far.

Aug 25 2020, 7:36 PM · Restricted Project
tinti updated subscribers of D85024: [llvm-objdump] Implement --prefix option.
Aug 25 2020, 7:34 PM · Restricted Project
tinti updated the diff for D85024: [llvm-objdump] Implement --prefix option.

Just adding my work so far.

Aug 25 2020, 9:16 AM · Restricted Project
tinti added inline comments to D85024: [llvm-objdump] Implement --prefix option.
Aug 25 2020, 9:13 AM · Restricted Project

Aug 24 2020

tinti added inline comments to D85024: [llvm-objdump] Implement --prefix option.
Aug 24 2020, 5:57 PM · Restricted Project
tinti updated the diff for D85024: [llvm-objdump] Implement --prefix option.

Use ;; for comments.
Remove double blanks.
Remove unsupported for Windows [1].
Add helper function of objdump::is_absolute().
Remove duplicated checks.
Rename TRAILING check.

Aug 24 2020, 5:56 PM · Restricted Project
tinti added inline comments to D85024: [llvm-objdump] Implement --prefix option.
Aug 24 2020, 2:22 PM · Restricted Project

Aug 16 2020

tinti added inline comments to D85024: [llvm-objdump] Implement --prefix option.
Aug 16 2020, 3:16 PM · Restricted Project
tinti updated the diff for D85024: [llvm-objdump] Implement --prefix option.

Fix documentation.

Aug 16 2020, 3:11 PM · Restricted Project

Aug 11 2020

tinti updated the diff for D85024: [llvm-objdump] Implement --prefix option.

Reword llvm/docs/CommandGuide/llvm-objdump.rst
Keep Windows as unsupported. [1]
Use -D in FileCheck for better testing.
Put primary test on top of the file. The primary use is to fix a wrong or incomplete path.
Fix missing --check-prefix on trailing test.
Remove comments about compatibility with GNU's objdump.
Remove namespace llvm:: where it is not needed.
Remove braces.
Add test comments.
Remove test checks that can be removed. [2]

Aug 11 2020, 7:32 PM · Restricted Project

Aug 10 2020

tinti added a comment to D85024: [llvm-objdump] Implement --prefix option.
Aug 10 2020, 4:47 AM · Restricted Project

Aug 6 2020

tinti updated the diff for D85024: [llvm-objdump] Implement --prefix option.

Add documentation.
Remove Windows test.
Implement remove trailing separators on prefix.
Add a real usage test case (wrong prefix fixed by --prefix).

Aug 6 2020, 7:13 PM · Restricted Project

Aug 4 2020

tinti added a comment to D85024: [llvm-objdump] Implement --prefix option.

Hi @tinti - please remember to upload with full diff context. I previously pointed you at the page on how to do so. Not including full context makes things harder to review.

Aug 4 2020, 6:32 PM · Restricted Project
tinti added inline comments to D85024: [llvm-objdump] Implement --prefix option.
Aug 4 2020, 4:02 AM · Restricted Project
tinti added inline comments to D85024: [llvm-objdump] Implement --prefix option.
Aug 4 2020, 3:56 AM · Restricted Project

Aug 3 2020

tinti updated the diff for D85024: [llvm-objdump] Implement --prefix option.
  • Add documentation.
  • Add a FIXME on Windows.
  • Fix a forced absolute paths bug on Unix-lie.
Aug 3 2020, 6:28 PM · Restricted Project
tinti added a comment to D85024: [llvm-objdump] Implement --prefix option.

Please remember to update the documentation located at llvm/docs/CommandGuide for llvm-objdump, to include the new option.

Aug 3 2020, 6:27 PM · Restricted Project

Aug 1 2020

tinti updated the diff for D85024: [llvm-objdump] Implement --prefix option.

Make old test unsupported for Windows.
Add new test for Windows.
Reword --prefix help text.

Aug 1 2020, 12:27 PM · Restricted Project

Jul 31 2020

tinti updated the diff for D85024: [llvm-objdump] Implement --prefix option.

Expected to fix failed test case on Windows.

Jul 31 2020, 6:55 PM · Restricted Project
tinti added inline comments to D85024: [llvm-objdump] Implement --prefix option.
Jul 31 2020, 5:33 PM · Restricted Project
tinti added inline comments to D85024: [llvm-objdump] Implement --prefix option.
Jul 31 2020, 9:02 AM · Restricted Project
tinti requested review of D85024: [llvm-objdump] Implement --prefix option.
Jul 31 2020, 5:27 AM · Restricted Project

Feb 11 2016

tinti awarded D16851: Update of "GCC extensions not implemented yet" in Clang User's Manual a Like token.
Feb 11 2016, 3:28 AM

Nov 20 2015

tinti closed D14804: [clang] Disable Unicode in asm files.

r253738

Nov 20 2015, 3:59 PM
tinti committed rL253738: [clang] Disable Unicode in asm files.
[clang] Disable Unicode in asm files
Nov 20 2015, 3:45 PM
tinti committed rL253737: Test commit.
Test commit
Nov 20 2015, 3:22 PM

Nov 18 2015

tinti set the repository for D14804: [clang] Disable Unicode in asm files to rL LLVM.
Nov 18 2015, 6:41 PM
tinti updated the diff for D14804: [clang] Disable Unicode in asm files.
Nov 18 2015, 6:41 PM
tinti added a reviewer for D14804: [clang] Disable Unicode in asm files: mehdi_amini.
Nov 18 2015, 6:39 PM
tinti retitled D14804: [clang] Disable Unicode in asm files from to [clang] Disable Unicode in asm files.
Nov 18 2015, 6:38 PM

Nov 6 2015

tinti added inline comments to D14184: [clang] Add initial support for -meabi flag.
Nov 6 2015, 2:55 PM
tinti updated D12413: [llvm] Add initial support for -meabi flag.
Nov 6 2015, 2:46 PM
tinti updated the diff for D12413: [llvm] Add initial support for -meabi flag.
Nov 6 2015, 2:45 PM
tinti updated the diff for D14184: [clang] Add initial support for -meabi flag.
  • Add test for error check
  • Change StringSwitch to use lllvm::EABI type
Nov 6 2015, 2:42 PM
tinti updated the diff for D14184: [clang] Add initial support for -meabi flag.
Nov 6 2015, 12:59 PM

Nov 5 2015

tinti added a comment to D12413: [llvm] Add initial support for -meabi flag.

Hi Vinicius,

This now looks good to me. Thanks!

You may have to change your clang patch, too, due to the change in -meabi here.

Nov 5 2015, 12:42 PM
tinti retitled D12413: [llvm] Add initial support for -meabi flag from [llvm] Add initial support for -eabi flag to [llvm] Add initial support for -meabi flag.
Nov 5 2015, 12:38 PM
tinti added a comment to D12413: [llvm] Add initial support for -meabi flag.
Nov 5 2015, 12:26 PM
tinti added a comment to D12413: [llvm] Add initial support for -meabi flag.
Nov 5 2015, 12:23 PM
tinti updated the diff for D12413: [llvm] Add initial support for -meabi flag.
  • Change flag to "meabi" to be keep consistence with Clang.
  • Use full call for memops functions to not be missinterpreted from memops functions.
  • Remove extra CHECK.
Nov 5 2015, 12:23 PM

Nov 4 2015

tinti added inline comments to D14184: [clang] Add initial support for -meabi flag.
Nov 4 2015, 6:27 PM
tinti updated the diff for D14184: [clang] Add initial support for -meabi flag.

Clang format.

Nov 4 2015, 6:24 PM
tinti added a comment to D12413: [llvm] Add initial support for -meabi flag.
Nov 4 2015, 6:23 PM
tinti added a comment to D12413: [llvm] Add initial support for -meabi flag.
Nov 4 2015, 6:23 PM
tinti updated the diff for D12413: [llvm] Add initial support for -meabi flag.
  • Use enum class.
  • Clang format (except in two places for readability).
Nov 4 2015, 6:21 PM
tinti added inline comments to D12413: [llvm] Add initial support for -meabi flag.
Nov 4 2015, 2:19 PM

Nov 3 2015

tinti added inline comments to D12413: [llvm] Add initial support for -meabi flag.
Nov 3 2015, 1:33 PM
tinti updated the diff for D14184: [clang] Add initial support for -meabi flag.
  • Clang format code.
  • Update eabi names to match new LLVM patch.
Nov 3 2015, 1:08 PM

Oct 30 2015

tinti added inline comments to D14184: [clang] Add initial support for -meabi flag.
Oct 30 2015, 10:33 AM
tinti added inline comments to D14184: [clang] Add initial support for -meabi flag.
Oct 30 2015, 10:24 AM
tinti updated the diff for D14184: [clang] Add initial support for -meabi flag.

Add context

Oct 30 2015, 6:31 AM
tinti updated the diff for D12413: [llvm] Add initial support for -meabi flag.

Add context

Oct 30 2015, 6:31 AM

Oct 29 2015

tinti updated subscribers of D14184: [clang] Add initial support for -meabi flag.
Oct 29 2015, 2:10 PM
tinti retitled D12413: [llvm] Add initial support for -meabi flag from [ARM] Add -arm-eabi for controlling RTLIB to [llvm] Add initial support for -eabi flag.
Oct 29 2015, 12:46 PM
tinti updated subscribers of D14184: [clang] Add initial support for -meabi flag.
Oct 29 2015, 12:44 PM
tinti retitled D14184: [clang] Add initial support for -meabi flag from to [clang] Add initial support for -meabi flag.
Oct 29 2015, 12:43 PM

Oct 27 2015

tinti added inline comments to D12413: [llvm] Add initial support for -meabi flag.
Oct 27 2015, 11:12 AM
tinti updated the diff for D12413: [llvm] Add initial support for -meabi flag.

Create -eabi flag to control EABI version.

Oct 27 2015, 11:09 AM

Oct 20 2015

tinti accepted D13792: Fix incorrect target triple in fp16-promote.ll.
Oct 20 2015, 1:43 PM
tinti added a comment to D13792: Fix incorrect target triple in fp16-promote.ll.

Frontends emit LLVM IR with --- (three dashes) to avoid ambiguous interpretation. Example:

Oct 20 2015, 1:42 PM
tinti added a comment to D13792: Fix incorrect target triple in fp16-promote.ll.

More precisely here:

Oct 20 2015, 1:04 PM
tinti added a comment to D13792: Fix incorrect target triple in fp16-promote.ll.

Hi,

This is weird, and I'd like to know more why the extra dash makes that big a difference...

Oct 20 2015, 12:35 PM

Sep 10 2015

tinti added inline comments to D12413: [llvm] Add initial support for -meabi flag.
Sep 10 2015, 4:23 PM

Sep 8 2015

tinti added inline comments to D12413: [llvm] Add initial support for -meabi flag.
Sep 8 2015, 3:27 PM
tinti added a comment to D12413: [llvm] Add initial support for -meabi flag.

Update according to @compnerd and @rengolin suggestions.

Sep 8 2015, 2:17 PM
tinti updated the diff for D12413: [llvm] Add initial support for -meabi flag.
Sep 8 2015, 2:16 PM

Sep 7 2015

tinti updated the diff for D12413: [llvm] Add initial support for -meabi flag.

Update patch and reword properly.

Sep 7 2015, 11:59 AM

Aug 31 2015

tinti added a comment to D12413: [llvm] Add initial support for -meabi flag.

I see. Well, clang_rt.builtins-arm.a does provide these already. Perhaps we should ask the GCC folks to consider moving the aliases from glibc into libgcc?

We're contemplating this one. Vinicius will prepare an email.

That seems to be where the disconnect is, it is with the user's explicit request. Using the "gnueabi" environment, *is* the request.

If "GNUEABI" meant something, that would be a request. But it doesn't.

Or rather, it means "whatever GCC does with the EABI", which is a pretty poor description, and even so, we're not doing "it".

There already is: use the armv7-linux-gnu environment, because you don't wish to conform to the EABI RTABI. You won't get the libcall functions "changed into other things". If this has an issue, we should absolutely fix that.

Right, this is yet another solution, and one that can be tested right now! Vinicius, does changing the triple helps the kernel?

Aug 31 2015, 12:24 PM

Aug 30 2015

tinti added a comment to D12413: [llvm] Add initial support for -meabi flag.

Related stuff:

Aug 30 2015, 10:00 AM
tinti added a comment to D12413: [llvm] Add initial support for -meabi flag.

The __aeabi_memcpy can just use memcpy. You can alias __aeabi_memcpy4 and __aeabi_memcpy8 to memcpy. The latter two are merely optimizations, and falling back to the unaligned use of memcpy is technically correct at a slight performance cost (which you would end up paying with your original suggestion anyways). Similarly, __aeabi_memset can use memset, and __aeabi_memmove can use memmove.

Aug 30 2015, 9:29 AM
tinti added a comment to D12413: [llvm] Add initial support for -meabi flag.

How do the ARM Compilers handle memcpy()?

Aug 30 2015, 4:22 AM

Aug 29 2015

tinti added a comment to D12413: [llvm] Add initial support for -meabi flag.
arm-linux-gnueabihf-ld -EL -p --no-undefined -X --pic-veneer --build-id -o .tmp_vmlinux1 -T ./arch/arm/kernel/vmlinux.lds arch/arm/kernel/head.o init/built-in.o --start-group usr/built-in.o arch/arm/vfp/built-in.o arch/arm/kernel/built-in.o arch/arm/mm/built-in.o arch/arm/common/built-in.o arch/arm/probes/built-in.o arch/arm/net/built-in.o arch/arm/crypto/built-in.o arch/arm/firmware/built-in.o arch/arm/mach-vexpress/built-in.o arch/arm/plat-versatile/built-in.o kernel/built-in.o mm/built-in.o fs/built-in.o ipc/built-in.o security/built-in.o crypto/built-in.o block/built-in.o arch/arm/lib/lib.a lib/lib.a arch/arm/lib/built-in.o lib/built-in.o drivers/built-in.o sound/built-in.o firmware/built-in.o arch/arm/oprofile/built-in.o net/built-in.o --end-group
arch/arm/kernel/built-in.o: In function `copy_thread':
./src/linux/arch/arm/kernel/process.c:214: undefined reference to `__aeabi_memcpy4'
Aug 29 2015, 12:08 PM
tinti added a comment to D12413: [llvm] Add initial support for -meabi flag.

I don't think that this is right. "gnueabi" isn't its own RT, but rather glibc + eabi (vs something like musleabi or androideabi). This is something which really should be fixed in glibc.

Ok. Does this mean that the __aeabi_memcpy and family should not be missing on glibc? As far as I saw gcc will never emit them.
Would not be acceptable for LLVM to not emit them while this does not get fixed?

GLIBC does provides __aebi_xxx symbols, just check if this build and runs fine:

Aug 29 2015, 6:52 AM

Aug 28 2015

tinti added a comment to D12413: [llvm] Add initial support for -meabi flag.

I don't think that this is right. "gnueabi" isn't its own RT, but rather glibc + eabi (vs something like musleabi or androideabi). This is something which really should be fixed in glibc.

Aug 28 2015, 12:57 PM

Aug 27 2015

tinti added a comment to D12413: [llvm] Add initial support for -meabi flag.

References used in this patch:

Aug 27 2015, 2:43 PM
tinti added reviewers for D12413: [llvm] Add initial support for -meabi flag: t.p.northover, compnerd.
Aug 27 2015, 2:20 PM
tinti set the repository for D12413: [llvm] Add initial support for -meabi flag to rL LLVM.
Aug 27 2015, 1:29 PM