grimar (George Rimar)
User

Projects

User does not belong to any projects.

User Details

User Since
Sep 21 2015, 12:36 AM (116 w, 2 d)

Recent Activity

Mon, Dec 11

grimar added a comment to D41046: [ELF] Make overlapping output sections an error.
In D41046#952097, @ruiu wrote:

If that's the case, creating overlapping sections is not an issue, but actually we have to be able to create such sections, no?

Mon, Dec 11, 10:14 PM
grimar added a comment to D41046: [ELF] Make overlapping output sections an error.
In D41046#952044, @ruiu wrote:

We do have a check if "." doesn't rewind. Doesn't it work as a prevention for section overlapping?

Mon, Dec 11, 9:48 PM
grimar added a comment to D41037: Create reserved symbols early so they can be versioned.

Sorry for delay, I can check and review that tomorrow (if Rui will not approve it until that).

Mon, Dec 11, 6:43 AM
grimar added inline comments to D41057: Avoid a gnu readelf warning.
Mon, Dec 11, 4:29 AM
grimar added a comment to D41046: [ELF] Make overlapping output sections an error.

Speed up lld invocation in test/ELF/many-alloc-sections.s lld invocation from 290s to <1s by not checking empty sections

Mon, Dec 11, 4:12 AM
grimar added a comment to D41041: [ELF] Don't set the executable bit for relocatable or shared files.

+1 to testcase request.

Mon, Dec 11, 3:44 AM
grimar added a comment to D39348: Implement --just-symbols..
In D39348#950733, @PkmX wrote:

If the only input file is specified as the argument of --just-symbol, lld will error out (ld.bfd accepts this):

$ riscv64-unknown-linux-gnu-ld.lld -r --just-symbols foo.o -o bar.o
riscv64-unknown-linux-gnu-ld.lld: error: no input files
riscv64-unknown-linux-gnu-ld.lld: error: target emulation unknown: -m or at least one .o file required
 

I'd suggest adding a separate kind of InputFile whose symbols only gets added to the symbol table as absolute (but don't add its sections). I have a patch that implements this if you'd like to review it.

Mon, Dec 11, 3:28 AM

Fri, Dec 8

grimar planned changes to D40954: [ELF] - Teach LLD to hint about -fdebug-types-section..

Thanks for comments, Rui. I'll update it after D40962 be landed (I think it will be and my patch intersects with it).

Fri, Dec 8, 1:54 AM
grimar added a comment to D40962: [ELF] When a relocation is out of range print the value and the range.

Patch LGTM, please wait for Rui approval.

Fri, Dec 8, 1:40 AM
grimar planned changes to D40950: [ELF] - Fail when multiple .debug_* sections are used in a single object..

This would not work for case when we have multiple .debug_type sections,
though it is legal and already supported. I'll update patch with the testcase for that
(if approach chosen by this patch be approved during current discussion in mail list).

Fri, Dec 8, 12:03 AM

Thu, Dec 7

grimar updated the diff for D40954: [ELF] - Teach LLD to hint about -fdebug-types-section..
  • Changed message format.
Thu, Dec 7, 7:34 AM
grimar planned changes to D40954: [ELF] - Teach LLD to hint about -fdebug-types-section..
Thu, Dec 7, 7:31 AM
grimar created D40954: [ELF] - Teach LLD to hint about -fdebug-types-section..
Thu, Dec 7, 7:27 AM
grimar updated the summary of D40950: [ELF] - Fail when multiple .debug_* sections are used in a single object..
Thu, Dec 7, 4:42 AM
grimar created D40950: [ELF] - Fail when multiple .debug_* sections are used in a single object..
Thu, Dec 7, 4:39 AM

Wed, Dec 6

grimar committed rLLD319879: [ELF] - Show .plt symbols in -Map output..
[ELF] - Show .plt symbols in -Map output.
Wed, Dec 6, 12:37 AM
grimar committed rL319879: [ELF] - Show .plt symbols in -Map output..
[ELF] - Show .plt symbols in -Map output.
Wed, Dec 6, 12:37 AM
grimar closed D40839: [ELF] - Show .plt symbols in -Map output. by committing rL319879: [ELF] - Show .plt symbols in -Map output..
Wed, Dec 6, 12:37 AM

Tue, Dec 5

grimar created D40839: [ELF] - Show .plt symbols in -Map output..
Tue, Dec 5, 8:14 AM
grimar added a comment to D36742: [LLD][ELF][AArch64] Implement scanner for Cortex-A53 Erratum 843419.

Looks you forgot to add AArch64ErrataFix.cpp.

Tue, Dec 5, 8:07 AM
grimar committed rL319769: Recommit r319747 "[ELF] - Show copy relocated symbols in a --Map file output.".
Recommit r319747 "[ELF] - Show copy relocated symbols in a --Map file output."
Tue, Dec 5, 6:24 AM
grimar committed rLLD319769: Recommit r319747 "[ELF] - Show copy relocated symbols in a --Map file output.".
Recommit r319747 "[ELF] - Show copy relocated symbols in a --Map file output."
Tue, Dec 5, 6:23 AM
grimar committed rL319752: Fix build bot after r319750 "[Support/TarWriter] - Don't allow TarWriter to add….
Fix build bot after r319750 "[Support/TarWriter] - Don't allow TarWriter to add…
Tue, Dec 5, 2:35 AM
grimar committed rL319750: [Support/TarWriter] - Don't allow TarWriter to add the same file more than once..
[Support/TarWriter] - Don't allow TarWriter to add the same file more than once.
Tue, Dec 5, 2:10 AM
grimar closed D40606: [Support/TarWriter] - Don't allow TarWriter to add the same file more than once. by committing rL319750: [Support/TarWriter] - Don't allow TarWriter to add the same file more than once..
Tue, Dec 5, 2:10 AM
grimar committed rLLD319749: [ELF] - Revert r319747 "[ELF] - Show copy relocated symbols in a --Map file….
[ELF] - Revert r319747 "[ELF] - Show copy relocated symbols in a --Map file…
Tue, Dec 5, 1:38 AM
grimar committed rL319749: [ELF] - Revert r319747 "[ELF] - Show copy relocated symbols in a --Map file….
[ELF] - Revert r319747 "[ELF] - Show copy relocated symbols in a --Map file…
Tue, Dec 5, 1:38 AM
grimar committed rLLD319747: [ELF] - Show copy relocated symbols in a --Map file output..
[ELF] - Show copy relocated symbols in a --Map file output.
Tue, Dec 5, 1:15 AM
grimar committed rL319747: [ELF] - Show copy relocated symbols in a --Map file output..
[ELF] - Show copy relocated symbols in a --Map file output.
Tue, Dec 5, 1:14 AM
grimar closed D40785: [ELF] - Show copy relocated symbols in a --Map file output. by committing rL319747: [ELF] - Show copy relocated symbols in a --Map file output..
Tue, Dec 5, 1:14 AM
grimar added a comment to D40785: [ELF] - Show copy relocated symbols in a --Map file output..
In D40785#943896, @ruiu wrote:

Actually I do have the same patch. Please take a look at https://reviews.llvm.org/D40630

Tue, Dec 5, 1:12 AM

Mon, Dec 4

grimar created D40785: [ELF] - Show copy relocated symbols in a --Map file output..
Mon, Dec 4, 7:02 AM
grimar updated the diff for D40606: [Support/TarWriter] - Don't allow TarWriter to add the same file more than once..

Addressed review comments:

  • Split test into three.
  • Use llvm::sys::fs::file_size API for finding output tar file size.
Mon, Dec 4, 12:20 AM

Sun, Dec 3

grimar abandoned D40549: [ELF] - Add support for --just-symbols flag..

Abdndoned in favor of D39348.

Sun, Dec 3, 11:09 PM

Fri, Dec 1

grimar abandoned D40558: [ELF] - Trigger error when -R <filename> is given..

Abandoning in according to discussions about this one in corresponding llvm-mails thread.

Fri, Dec 1, 11:32 PM
grimar abandoned D39047: [ELF] - Hide LinkerScript::removeEmptyCommands()..
Fri, Dec 1, 4:19 AM
grimar abandoned D38582: [ELF] - Get rid of LinkerScript::adjustSectionsBeforeSorting()..
Fri, Dec 1, 4:18 AM
grimar abandoned D33488: [ELF] - Optimization for populating stringpool when building .gdb_index..

Irrelevant.

Fri, Dec 1, 4:16 AM
grimar abandoned D40128: [ELF] - Reveal layout of synthetic mergeable sections when producing -Map.
Fri, Dec 1, 4:16 AM
grimar abandoned D38036: [yaml2obj] - Don't crash on invalid document..

Not sure it was the best approach possible. Abandoning.

Fri, Dec 1, 4:15 AM
grimar abandoned D35945: [ELF] - Linkerscript: better diagnostic for INPUT/GROUP commands..
Fri, Dec 1, 4:14 AM
grimar abandoned D39489: [ELF] - Linkerscript: fix issue with multiple output sections definitions..
Fri, Dec 1, 4:14 AM
grimar abandoned D37013: [ELF] - Add additional information about location when emiting linkerscript's parsing errors..
Fri, Dec 1, 4:13 AM
grimar abandoned D37520: [ELF] - Fix removing of unused synthetic sections..
Fri, Dec 1, 4:13 AM
grimar abandoned D39096: [ELF] - Generalize finding output section by name..
Fri, Dec 1, 4:13 AM
grimar abandoned D39418: [ELF] - Split processSectionCommands()..
Fri, Dec 1, 4:12 AM
grimar abandoned D38180: [ELF] - ICF: improve support of SHF_LINK_ORDER sections..
Fri, Dec 1, 4:12 AM
grimar updated the diff for D40606: [Support/TarWriter] - Don't allow TarWriter to add the same file more than once..
  • Added testcase.
Fri, Dec 1, 3:23 AM
grimar committed rL319527: [ELF] - Rename excessive variable. NFC..
[ELF] - Rename excessive variable. NFC.
Fri, Dec 1, 1:15 AM
grimar committed rLLD319527: [ELF] - Rename excessive variable. NFC..
[ELF] - Rename excessive variable. NFC.
Fri, Dec 1, 1:15 AM
grimar committed rLLD319526: [ELF] - Produce relocation section name consistent with output section name….
[ELF] - Produce relocation section name consistent with output section name…
Fri, Dec 1, 1:05 AM
grimar committed rL319526: [ELF] - Produce relocation section name consistent with output section name….
[ELF] - Produce relocation section name consistent with output section name…
Fri, Dec 1, 1:05 AM
grimar closed D40652: [ELF] - Produce relocation section name consistent with output section name when --emit-reloc used with linker script. by committing rL319526: [ELF] - Produce relocation section name consistent with output section name….
Fri, Dec 1, 1:05 AM
grimar added inline comments to D40652: [ELF] - Produce relocation section name consistent with output section name when --emit-reloc used with linker script..
Fri, Dec 1, 12:59 AM
grimar closed D40661: Error on -no-define-common if the output is not relocatable.

Was committed as r319473.

Fri, Dec 1, 12:45 AM
grimar accepted D40661: Error on -no-define-common if the output is not relocatable.

Committed version LGTM too.

Fri, Dec 1, 12:45 AM

Thu, Nov 30

grimar updated the diff for D40606: [Support/TarWriter] - Don't allow TarWriter to add the same file more than once..
  • Use llvm::StringSet instead of std::set<std::string>.
Thu, Nov 30, 9:44 AM
grimar created D40652: [ELF] - Produce relocation section name consistent with output section name when --emit-reloc used with linker script..
Thu, Nov 30, 7:28 AM
grimar committed rLLD319435: [ELF] - Handle EhInputSection Live bit in MarkLive.cpp.
[ELF] - Handle EhInputSection Live bit in MarkLive.cpp
Thu, Nov 30, 6:02 AM
grimar committed rL319435: [ELF] - Handle EhInputSection Live bit in MarkLive.cpp.
[ELF] - Handle EhInputSection Live bit in MarkLive.cpp
Thu, Nov 30, 6:01 AM
grimar closed D40454: [ELF] - Handle EhInputSection Live bit in MarkLive.cpp by committing rL319435: [ELF] - Handle EhInputSection Live bit in MarkLive.cpp.
Thu, Nov 30, 6:01 AM
grimar added inline comments to D40606: [Support/TarWriter] - Don't allow TarWriter to add the same file more than once..
Thu, Nov 30, 3:08 AM
grimar updated the diff for D40606: [Support/TarWriter] - Don't allow TarWriter to add the same file more than once..
  • Addressed review comments.
Thu, Nov 30, 3:05 AM

Wed, Nov 29

grimar created D40606: [Support/TarWriter] - Don't allow TarWriter to add the same file more than once..
Wed, Nov 29, 8:00 AM

Tue, Nov 28

grimar added a comment to D40549: [ELF] - Add support for --just-symbols flag..

FYI @ruiu has already wrote a similar patch in D39348.

Tue, Nov 28, 11:55 PM
grimar created D40558: [ELF] - Trigger error when -R <filename> is given..
Tue, Nov 28, 7:16 AM
grimar created D40549: [ELF] - Add support for --just-symbols flag..
Tue, Nov 28, 4:56 AM

Mon, Nov 27

grimar created D40484: [DEMO].
Mon, Nov 27, 4:35 AM
grimar updated subscribers of D40484: [DEMO].
Mon, Nov 27, 4:35 AM
grimar updated the diff for D40352: [DEMO].
Mon, Nov 27, 4:30 AM

Sat, Nov 25

grimar created D40454: [ELF] - Handle EhInputSection Live bit in MarkLive.cpp.
Sat, Nov 25, 2:04 AM

Thu, Nov 23

grimar added inline comments to D40365: [ELF] Rename .bss.rel.ro to .data.rel.ro.bss for some Linker Scripts..
Thu, Nov 23, 10:24 PM
grimar updated the diff for D40352: [DEMO].
Thu, Nov 23, 7:02 AM
grimar added a comment to D40170: [llvm-tblgen] - Stop using std::string in RecordKeeper..

BTW, there was a delay of 24h between commit and revert of this and no bots reported any failtures to me. I wonder why it happened ? Don't we have ones with memory sanitizer enabled ?
(I did not test it with -DLLVM_USE_SANITIZER=Memory yet, but just wonder how it is possible ?).

Thu, Nov 23, 2:32 AM

Wed, Nov 22

grimar added a comment to D40170: [llvm-tblgen] - Stop using std::string in RecordKeeper..

When built with memory sanitizer, this change causes many use-of-uninitialized-value errors and also causes compilation failures when built with the cmake option -DBUILD_SHARED_LIBS=ON. AFAIK, StringRef does not store the string contents in itself and just maintains a pointer to an external storage. Are you sure all the string keys supplied to Classes or Defs map in RecordKeeper class are persistent throughout all the uses of those maps?

If you can't reproduce the same error, try building with the memory sanitizer, which can be enabled with the cmake option -DLLVM_USE_SANITIZER=Memory. If you confirm this problem, I suggest maybe reverting this change.

Wed, Nov 22, 10:57 PM
grimar committed rL318899: Revert r318822 "[llvm-tblgen] - Stop using std::string in RecordKeeper.".
Revert r318822 "[llvm-tblgen] - Stop using std::string in RecordKeeper."
Wed, Nov 22, 10:53 PM
grimar created D40352: [DEMO].
Wed, Nov 22, 5:38 AM

Tue, Nov 21

grimar committed rL318822: [llvm-tblgen] - Stop using std::string in RecordKeeper..
[llvm-tblgen] - Stop using std::string in RecordKeeper.
Tue, Nov 21, 11:55 PM
grimar closed D40170: [llvm-tblgen] - Stop using std::string in RecordKeeper. by committing rL318822: [llvm-tblgen] - Stop using std::string in RecordKeeper..
Tue, Nov 21, 11:55 PM

Mon, Nov 20

grimar updated the diff for D40170: [llvm-tblgen] - Stop using std::string in RecordKeeper..
  • Use StringRef in Tag instead of std::string*'. (Tag` remembers ID of Classes/Defs from Record and since patch changes type of map key from std::string to StringRef, this place should be updated.)
Mon, Nov 20, 6:00 AM
grimar planned changes to D40170: [llvm-tblgen] - Stop using std::string in RecordKeeper..
Mon, Nov 20, 3:54 AM
grimar added inline comments to D39493: [ELF] Fix DT_MIPS_LOCAL_GOTNO value when using linker scripts to change section sizes.
Mon, Nov 20, 3:12 AM
grimar abandoned D40239: [DEMO].
Mon, Nov 20, 1:15 AM
grimar added a comment to D40170: [llvm-tblgen] - Stop using std::string in RecordKeeper..

I just realized that I do not know if ordering is important or not. I know almost nothing about TableGen.
Some code iterates over std::map returned, but it does not automatially means ordering is really important.

Mon, Nov 20, 1:15 AM
grimar added a comment to D40170: [llvm-tblgen] - Stop using std::string in RecordKeeper..

As for this patch, I would actually prefer if we didn't even use std::map. Is ordering important here? Can this be changed to a llvm::StringMap?

Mon, Nov 20, 12:44 AM
grimar created D40239: [DEMO].
Mon, Nov 20, 12:36 AM

Fri, Nov 17

grimar added a comment to D40129: [ELF] Fall back to search dirs for linker scripts specified with -T.

(please wait approvement from Rui)

Fri, Nov 17, 6:59 AM
grimar accepted D40129: [ELF] Fall back to search dirs for linker scripts specified with -T.

I have no futher comments, thanks !

Fri, Nov 17, 6:53 AM
grimar committed rL318519: [ELF] - Teach LLD to use information from .debug_str for error reporting..
[ELF] - Teach LLD to use information from .debug_str for error reporting.
Fri, Nov 17, 3:57 AM
grimar closed D39542: [ELF] - Teach LLD to use information from .debug_str for error reporting. by committing rL318519: [ELF] - Teach LLD to use information from .debug_str for error reporting..
Fri, Nov 17, 3:57 AM
grimar abandoned D39959: [ELF] - Allow merging of strings sections for -relocatable output..

Abandoning. Another approach was chosen (D40026) and landed (r318516 + r318518)

Fri, Nov 17, 3:47 AM
grimar committed rL318518: [ELF] - Fix BB after r318516.
[ELF] - Fix BB after r318516
Fri, Nov 17, 3:40 AM
grimar committed rL318516: [ELF] - Allow applying SHF_MERGE optimization for relocatable output..
[ELF] - Allow applying SHF_MERGE optimization for relocatable output.
Fri, Nov 17, 3:28 AM
grimar closed D40026: [ELF] - Allow applying SHF_MERGE optimization for relocatable output. by committing rL318516: [ELF] - Allow applying SHF_MERGE optimization for relocatable output..
Fri, Nov 17, 3:28 AM
grimar updated the summary of D40026: [ELF] - Allow applying SHF_MERGE optimization for relocatable output..
Fri, Nov 17, 3:27 AM
grimar added a comment to D40170: [llvm-tblgen] - Stop using std::string in RecordKeeper..

LLVM_OPTIMIZED_TABLEGEN speedups build more than 2x for me, I wonder why it is not default option. (we probably could emit warning or something that if no release binaries found to use).

Fri, Nov 17, 3:23 AM
grimar created D40170: [llvm-tblgen] - Stop using std::string in RecordKeeper..
Fri, Nov 17, 2:09 AM
grimar added a comment to D40128: [ELF] - Reveal layout of synthetic mergeable sections when producing -Map.
In D40128#928354, @ruiu wrote:

First of all, the current output is consistent. It doesn't print out the information about mergeable input sections that consist make up an mergeable synthetic section, but that doesn't mean that the output is inconsistent.

So, back to the original question, I don't think the format you are proposing is useful for a few reasons:

  • Address and size fields contains bogus values. So you cannot use these numbers to analyze the output
Fri, Nov 17, 1:03 AM
grimar added a comment to D40129: [ELF] Fall back to search dirs for linker scripts specified with -T.

Thanks for this patch, few comments/suggestions below.

Fri, Nov 17, 1:00 AM
grimar added a comment to D40128: [ELF] - Reveal layout of synthetic mergeable sections when producing -Map.
In D40128#928250, @ruiu wrote:

How is this useful?

Fri, Nov 17, 12:25 AM