Page MenuHomePhabricator

int3 (Jez Ng)
User

Projects

User does not belong to any projects.

User Details

User Since
Jan 22 2020, 2:12 PM (73 w, 2 d)

Recent Activity

Today

int3 accepted D104596: [lld/mac] Make sure __thread_ptrs is in front of __thread_bss.

Thanks! I see why you were checking for the GOT in tlv-dylib.s now...

Sat, Jun 19, 9:42 AM · Restricted Project, Restricted Project
int3 accepted D104594: [lld/mac] Give __DATA,__thread_ptrs type S_THREAD_LOCAL_VARIABLE_POINTERS.
Sat, Jun 19, 9:41 AM · Restricted Project, Restricted Project

Yesterday

int3 committed rG4507f64165fd: [re-land][lld-macho] Avoid force-loading the same archive twice (authored by int3).
[re-land][lld-macho] Avoid force-loading the same archive twice
Fri, Jun 18, 7:45 PM
int3 added a reverting change for rGc9b241efd68c: Revert "[lld-macho] Avoid force-loading the same archive twice": rG4507f64165fd: [re-land][lld-macho] Avoid force-loading the same archive twice.
Fri, Jun 18, 7:45 PM
int3 added a reverting change for rG1d31fb8d122b: [lld-macho] Have path-related functions return std::string, not StringRef: rGa79c01832553: Revert "[lld-macho] Have path-related functions return std::string, not….
Fri, Jun 18, 7:45 PM
int3 committed rGa79c01832553: Revert "[lld-macho] Have path-related functions return std::string, not… (authored by int3).
Revert "[lld-macho] Have path-related functions return std::string, not…
Fri, Jun 18, 7:45 PM
int3 added a comment to D104353: [lld-macho] Avoid force-loading the same archive twice.

The real culprit is https://github.com/llvm/llvm-project/commit/1d31fb8d122b1117cf20a9edc09812db8472e930, which I (unfortunately) thought simple enough to be landed without review. And I only landed it today, which explains why the breakage only happened today.

Fri, Jun 18, 7:27 PM · Restricted Project, Restricted Project
int3 committed rG1d31fb8d122b: [lld-macho] Have path-related functions return std::string, not StringRef (authored by int3).
[lld-macho] Have path-related functions return std::string, not StringRef
Fri, Jun 18, 1:36 PM
int3 committed rG4c49f9ceafd6: [lld-macho] Handle non-extern symbols marked as private extern (authored by int3).
[lld-macho] Handle non-extern symbols marked as private extern
Fri, Jun 18, 1:36 PM
int3 closed D104502: [lld-macho] Handle non-extern symbols marked as private extern.
Fri, Jun 18, 1:36 PM · Restricted Project, Restricted Project

Thu, Jun 17

int3 updated the diff for D104502: [lld-macho] Handle non-extern symbols marked as private extern.
Thu, Jun 17, 7:29 PM · Restricted Project, Restricted Project
int3 updated the diff for D104502: [lld-macho] Handle non-extern symbols marked as private extern.

remove REQUIRES

Thu, Jun 17, 7:06 PM · Restricted Project, Restricted Project
int3 requested review of D104502: [lld-macho] Handle non-extern symbols marked as private extern.
Thu, Jun 17, 6:45 PM · Restricted Project, Restricted Project
int3 accepted D103971: [lld-macho] Rework mergeFlag to closer mimic what LD64 does..

Thanks!

Thu, Jun 17, 10:51 AM · Restricted Project, Restricted Project
int3 added inline comments to D103971: [lld-macho] Rework mergeFlag to closer mimic what LD64 does..
Thu, Jun 17, 9:22 AM · Restricted Project, Restricted Project
int3 committed rG24706cd73cd1: [lld-macho] Avoid force-loading the same archive twice (authored by int3).
[lld-macho] Avoid force-loading the same archive twice
Thu, Jun 17, 8:14 AM
int3 closed D104353: [lld-macho] Avoid force-loading the same archive twice.
Thu, Jun 17, 8:14 AM · Restricted Project, Restricted Project
int3 added inline comments to D104353: [lld-macho] Avoid force-loading the same archive twice.
Thu, Jun 17, 8:12 AM · Restricted Project, Restricted Project
int3 added a comment to D103971: [lld-macho] Rework mergeFlag to closer mimic what LD64 does..

Sorry for the delay, was OOO earlier this week for a surgery.

Thu, Jun 17, 8:11 AM · Restricted Project, Restricted Project

Wed, Jun 16

int3 added a comment to D103971: [lld-macho] Rework mergeFlag to closer mimic what LD64 does..

Bump. Are you going to finish this? I've stumbled onto some programs that need it :)

Wed, Jun 16, 6:57 PM · Restricted Project, Restricted Project
int3 updated the diff for D104353: [lld-macho] Avoid force-loading the same archive twice.

remove use of ln; add test for force-loading multiple non-conflicting archives, and a test for non-force-loading multiple conflicting archives

Wed, Jun 16, 6:24 PM · Restricted Project, Restricted Project
int3 added inline comments to D104352: [lld-macho] Handle multiple LC_LINKER_OPTIONs.
Wed, Jun 16, 6:03 PM · Restricted Project, Restricted Project
int3 added inline comments to D104353: [lld-macho] Avoid force-loading the same archive twice.
Wed, Jun 16, 5:51 PM · Restricted Project, Restricted Project
int3 added inline comments to D104353: [lld-macho] Avoid force-loading the same archive twice.
Wed, Jun 16, 2:36 PM · Restricted Project, Restricted Project
int3 updated the diff for D104353: [lld-macho] Avoid force-loading the same archive twice.

test symlinks and hardlinks

Wed, Jun 16, 1:42 PM · Restricted Project, Restricted Project
int3 added a comment to D104353: [lld-macho] Avoid force-loading the same archive twice.

Yeah as mentioned in my inline comment, my experiments show that ld64 treats symlinks as distinct inputs. But @thakis seems to think its otherwise, so we'll see...

Wed, Jun 16, 1:42 PM · Restricted Project, Restricted Project
int3 committed rG560636e5497a: [lld-macho] Put DATA_IN_CODE immediately after FUNCTION_STARTS (authored by int3).
[lld-macho] Put DATA_IN_CODE immediately after FUNCTION_STARTS
Wed, Jun 16, 12:23 PM
int3 committed rGeeac6b2becee: [lld-macho] Handle multiple LC_LINKER_OPTIONs (authored by int3).
[lld-macho] Handle multiple LC_LINKER_OPTIONs
Wed, Jun 16, 12:23 PM
int3 committed rGb8bbb9723af3: [lld-macho][nfc] Put back shouldOmitFromOutput() asserts (authored by int3).
[lld-macho][nfc] Put back shouldOmitFromOutput() asserts
Wed, Jun 16, 12:23 PM
int3 closed D104354: [lld-macho] Put DATA_IN_CODE immediately after FUNCTION_STARTS.
Wed, Jun 16, 12:23 PM · Restricted Project, Restricted Project
int3 closed D104352: [lld-macho] Handle multiple LC_LINKER_OPTIONs.
Wed, Jun 16, 12:23 PM · Restricted Project, Restricted Project
int3 closed D104169: [lld-macho][nfc] Put back shouldOmitFromOutput() asserts.
Wed, Jun 16, 12:23 PM · Restricted Project, Restricted Project
int3 added inline comments to D104353: [lld-macho] Avoid force-loading the same archive twice.
Wed, Jun 16, 12:21 PM · Restricted Project, Restricted Project
int3 accepted D104346: [lld][MachO] Sort symbols in parallel in -map.

lgtm, thanks!

Wed, Jun 16, 11:52 AM · Restricted Project, Restricted Project, lld
int3 added inline comments to D104352: [lld-macho] Handle multiple LC_LINKER_OPTIONs.
Wed, Jun 16, 11:52 AM · Restricted Project, Restricted Project
int3 committed rGd52d1b93c3f5: [lld-macho] Downgrade version mismatch to warning (authored by int3).
[lld-macho] Downgrade version mismatch to warning
Wed, Jun 16, 8:06 AM
int3 closed D104333: [lld-macho] Downgrade version mismatch to warning.
Wed, Jun 16, 8:06 AM · Restricted Project, Restricted Project

Tue, Jun 15

int3 added inline comments to D104346: [lld][MachO] Sort symbols in parallel in -map.
Tue, Jun 15, 11:22 PM · Restricted Project, Restricted Project, lld
int3 updated the diff for D104353: [lld-macho] Avoid force-loading the same archive twice.

tweak test

Tue, Jun 15, 11:06 PM · Restricted Project, Restricted Project
int3 requested review of D104354: [lld-macho] Put DATA_IN_CODE immediately after FUNCTION_STARTS.
Tue, Jun 15, 10:11 PM · Restricted Project, Restricted Project
int3 added a comment to D104346: [lld][MachO] Sort symbols in parallel in -map.

Trying to produce one

Tue, Jun 15, 10:01 PM · Restricted Project, Restricted Project, lld
int3 accepted D104345: [lld/mac] Add support for -no_data_in_code_info flag.

lgtm

Tue, Jun 15, 9:54 PM · Restricted Project, Restricted Project
int3 accepted D104347: [lld/mac] Make sure all lld-specific flags show up in the "LLD-SPECIFIC:" --help section.

Thanks!

Tue, Jun 15, 9:54 PM · Restricted Project, Restricted Project
int3 requested review of D104353: [lld-macho] Avoid force-loading the same archive twice.
Tue, Jun 15, 9:54 PM · Restricted Project, Restricted Project
int3 requested review of D104352: [lld-macho] Handle multiple LC_LINKER_OPTIONs.
Tue, Jun 15, 9:53 PM · Restricted Project, Restricted Project
int3 updated the diff for D104333: [lld-macho] Downgrade version mismatch to warning.

update test

Tue, Jun 15, 9:53 PM · Restricted Project, Restricted Project
int3 added a comment to D104333: [lld-macho] Downgrade version mismatch to warning.

We did have a test actually. I just I forgot we had %no_fatal_warnings_lld, which makes it not-too-ugly to test for warnings instead of errors... updating that test now

Tue, Jun 15, 9:53 PM · Restricted Project, Restricted Project
int3 requested review of D104333: [lld-macho] Downgrade version mismatch to warning.
Tue, Jun 15, 4:22 PM · Restricted Project, Restricted Project

Mon, Jun 14

int3 added inline comments to D104169: [lld-macho][nfc] Put back shouldOmitFromOutput() asserts.
Mon, Jun 14, 11:00 AM · Restricted Project, Restricted Project
int3 updated the diff for D104169: [lld-macho][nfc] Put back shouldOmitFromOutput() asserts.

use isCoalescedWeak

Mon, Jun 14, 11:00 AM · Restricted Project, Restricted Project
int3 committed rGcc17bfe4898d: [lld-macho] Fix "shift exponent too large" UBSAN error (authored by int3).
[lld-macho] Fix "shift exponent too large" UBSAN error
Mon, Jun 14, 10:47 AM
int3 committed rGe06b9ba48562: [lld-macho] Reword comment for clarity (authored by int3).
[lld-macho] Reword comment for clarity
Mon, Jun 14, 10:47 AM

Sun, Jun 13

int3 committed rG9c5d43fb5513: [lld-macho] Try to fix MSAN "uninitialized memory" error (authored by int3).
[lld-macho] Try to fix MSAN "uninitialized memory" error
Sun, Jun 13, 8:47 PM
int3 committed rGda24e6d43e3f: [lld-macho][nfc] Add `final` to classes where possible (authored by int3).
[lld-macho][nfc] Add `final` to classes where possible
Sun, Jun 13, 4:52 PM
int3 committed rGc5c05ffa4562: [lld-macho][nfc] Represent the image loader cache with a ConcatInputSection (authored by int3).
[lld-macho][nfc] Represent the image loader cache with a ConcatInputSection
Sun, Jun 13, 4:52 PM
int3 committed rGb2a073901216: [lld-macho][nfc] Remove InputSection::outSecFileOff (authored by int3).
[lld-macho][nfc] Remove InputSection::outSecFileOff
Sun, Jun 13, 4:52 PM
int3 closed D104200: [lld-macho][nfc] Add `final` to classes where possible.
Sun, Jun 13, 4:52 PM · Restricted Project, Restricted Project
int3 closed D104199: [lld-macho][nfc] Represent the image loader cache with a ConcatInputSection.
Sun, Jun 13, 4:52 PM · Restricted Project, Restricted Project
int3 closed D104177: [lld-macho][nfc] Remove InputSection::outSecFileOff.
Sun, Jun 13, 4:52 PM · Restricted Project, Restricted Project
int3 updated the summary of D104177: [lld-macho][nfc] Remove InputSection::outSecFileOff.
Sun, Jun 13, 4:41 PM · Restricted Project, Restricted Project
int3 added a comment to D104177: [lld-macho][nfc] Remove InputSection::outSecFileOff.

Sorry, I think that was spurious :( I can no longer replicate that performance win...

Sun, Jun 13, 4:40 PM · Restricted Project, Restricted Project
int3 requested review of D104200: [lld-macho][nfc] Add `final` to classes where possible.
Sun, Jun 13, 10:55 AM · Restricted Project, Restricted Project
int3 retitled D104169: [lld-macho][nfc] Put back shouldOmitFromOutput() asserts from [lld-macho] Put back shouldOmitFromOutput() asserts to [lld-macho][nfc] Put back shouldOmitFromOutput() asserts.
Sun, Jun 13, 10:55 AM · Restricted Project, Restricted Project
int3 requested review of D104199: [lld-macho][nfc] Represent the image loader cache with a ConcatInputSection.
Sun, Jun 13, 10:55 AM · Restricted Project, Restricted Project
int3 retitled D104177: [lld-macho][nfc] Remove InputSection::outSecFileOff from [lld-macho] Remove InputSection::outSecFileOff to [lld-macho][nfc] Remove InputSection::outSecFileOff.
Sun, Jun 13, 10:55 AM · Restricted Project, Restricted Project

Fri, Jun 11

int3 requested review of D104177: [lld-macho][nfc] Remove InputSection::outSecFileOff.
Fri, Jun 11, 11:38 PM · Restricted Project, Restricted Project
int3 abandoned D104158: [not for review] Dump counts of literals/subsections.
Fri, Jun 11, 10:04 PM · Restricted Project, Restricted Project
int3 abandoned D104159: [not for review][lld-macho] Simple cstring literal implementation.
Fri, Jun 11, 10:03 PM · Restricted Project, Restricted Project
int3 requested review of D104169: [lld-macho][nfc] Put back shouldOmitFromOutput() asserts.
Fri, Jun 11, 7:01 PM · Restricted Project, Restricted Project
int3 added a comment to D103977: [lld-macho][nfc] Move liveness-tracking fields into ConcatInputSection.

I ended up removing a lot of the asserts since sprinkling casts to ConcatInputSection seemed pretty awkward. I think you added most of these asserts, so lmk if you think we should put them back

Fri, Jun 11, 5:32 PM · Restricted Project, Restricted Project
int3 committed rG5de7467e9821: [lld-macho] Fix debug build (authored by int3).
[lld-macho] Fix debug build
Fri, Jun 11, 5:21 PM
int3 added a comment to D103977: [lld-macho][nfc] Move liveness-tracking fields into ConcatInputSection.

And consider spreading out landing commits so that it's easier to see which one breaks things :)

Fri, Jun 11, 5:01 PM · Restricted Project, Restricted Project
int3 committed rG464d3dc3d11e: [lld-macho] Have dead-stripping work with literal sections (authored by int3).
[lld-macho] Have dead-stripping work with literal sections
Fri, Jun 11, 4:51 PM
int3 committed rG681cfeb59119: [lld-macho][nfc] Have InputSection ctors take some parameters (authored by int3).
[lld-macho][nfc] Have InputSection ctors take some parameters
Fri, Jun 11, 4:50 PM
int3 committed rG5d88f2dd9478: [lld-macho] Deduplicate fixed-width literals (authored by int3).
[lld-macho] Deduplicate fixed-width literals
Fri, Jun 11, 4:50 PM
int3 committed rG7f2ba39b1688: [lld-macho][nfc] Move liveness-tracking fields into ConcatInputSection (authored by int3).
[lld-macho][nfc] Move liveness-tracking fields into ConcatInputSection
Fri, Jun 11, 4:50 PM
int3 closed D103979: [lld-macho] Have dead-stripping work with literal sections.
Fri, Jun 11, 4:50 PM · Restricted Project, Restricted Project
int3 closed D103978: [lld-macho][nfc] Have InputSection ctors take some parameters.
Fri, Jun 11, 4:50 PM · Restricted Project, Restricted Project
int3 closed D103977: [lld-macho][nfc] Move liveness-tracking fields into ConcatInputSection.
Fri, Jun 11, 4:50 PM · Restricted Project, Restricted Project
int3 closed D103113: [lld-macho] Deduplicate fixed-width literals.
Fri, Jun 11, 4:50 PM · Restricted Project, Restricted Project
int3 added a comment to D103977: [lld-macho][nfc] Move liveness-tracking fields into ConcatInputSection.

But overall, it feels like things get a lot more complicated because we're not creating real InputSections for each literal in literal sections.

Fri, Jun 11, 4:38 PM · Restricted Project, Restricted Project
int3 requested review of D104159: [not for review][lld-macho] Simple cstring literal implementation.
Fri, Jun 11, 4:33 PM · Restricted Project, Restricted Project
int3 requested review of D104158: [not for review] Dump counts of literals/subsections.
Fri, Jun 11, 4:32 PM · Restricted Project, Restricted Project
int3 accepted D103292: [lld-macho] Implement ICF.

Neat! I think we still need a test for N_ALT_ENTRY, but I'm fine with having that in a follow-up (maybe just leave a TODO in icf.s).

Fri, Jun 11, 2:27 PM · Restricted Project, Restricted Project
int3 added a comment to D103971: [lld-macho] Rework mergeFlag to closer mimic what LD64 does..

It makes sense, yes. Always good to cross-check it with what ld64 is doing though.

Fri, Jun 11, 2:01 PM · Restricted Project, Restricted Project
int3 added a comment to D104084: [lld-macho] Add options -(no_)keep_dwarf_unwind.

Yes, I will do it, but wanted this out quickly.

Fri, Jun 11, 10:21 AM · Restricted Project, Restricted Project

Thu, Jun 10

int3 accepted D104081: [lld/mac] Make binaries written by lld strippable.

Thanks for the fix!

Thu, Jun 10, 8:18 PM · Restricted Project, Restricted Project
int3 added inline comments to D103292: [lld-macho] Implement ICF.
Thu, Jun 10, 3:35 PM · Restricted Project, Restricted Project
int3 added a comment to D103292: [lld-macho] Implement ICF.

Nice work :)

Thu, Jun 10, 2:57 PM · Restricted Project, Restricted Project
int3 accepted D104027: [lld/mac] Use sectionType() more.

Thanks!

Thu, Jun 10, 12:13 PM · Restricted Project, Restricted Project
int3 committed rG4b5c6c5c4b92: [lld-macho][nfc] Fix uninitialized members warning from Coverity (authored by int3).
[lld-macho][nfc] Fix uninitialized members warning from Coverity
Thu, Jun 10, 12:09 PM

Wed, Jun 9

int3 updated the diff for D103979: [lld-macho] Have dead-stripping work with literal sections.

static_assert

Wed, Jun 9, 10:10 PM · Restricted Project, Restricted Project
int3 added inline comments to D103979: [lld-macho] Have dead-stripping work with literal sections.
Wed, Jun 9, 10:02 PM · Restricted Project, Restricted Project
int3 added inline comments to D103979: [lld-macho] Have dead-stripping work with literal sections.
Wed, Jun 9, 8:49 PM · Restricted Project, Restricted Project
int3 accepted D103998: [lld/mac] Make framework symlinks in tests more realistic.

Realism is always good :)

Wed, Jun 9, 5:37 PM · Restricted Project, Restricted Project
int3 accepted D103990: [lld/mac] When handling @loader_path, use realpath() of symlinks.
Wed, Jun 9, 5:25 PM · Restricted Project, Restricted Project
int3 added a comment to D103292: [lld-macho] Implement ICF.

Ah I see. Fair enough, there seems no getting around that. Having a pointer to the canonical InputSection instead of a symbol would solve the problem with N_ALT_ENTRY mentioned earlier, though

Wed, Jun 9, 4:29 PM · Restricted Project, Restricted Project
int3 added inline comments to D103990: [lld/mac] When handling @loader_path, use realpath() of symlinks.
Wed, Jun 9, 4:28 PM · Restricted Project, Restricted Project
int3 added inline comments to D103985: [lld/mac] Print dylib search details with --print-dylib-search or RC_TRACE_DYLIB_SEARCHING.
Wed, Jun 9, 4:25 PM · Restricted Project, Restricted Project
int3 accepted D103985: [lld/mac] Print dylib search details with --print-dylib-search or RC_TRACE_DYLIB_SEARCHING.
Wed, Jun 9, 2:03 PM · Restricted Project, Restricted Project