Moved five of the DIFlags to DISPFlags. (updated few clang testcases)
DIFlagExplicit -> DISPFlagExplicit
DIFlagPrototyped -> DISPFlagPrototyped
DIFlagNoReturn -> DISPFlagNoReturn
DIFlagThunk -> DISPFlagThunk
DIFlagAllCallsDescribed -> DISPFlagAllCallsDescribed

In D74470#1872492, @probinson wrote:

D73261 would also benefit from moving more flags into the subprogram flags field. The refactoring really should move out as many flags as possible all in one go. I understand this is more than you anticipated but it will be the right thing to do for the project as a whole.

• some typos cleanup.
• updated bitcode test(both .bc and .ll files) to use old flags and checked against new btflags.

In D73861#1872481, @probinson wrote:

In D73861#1871591, @Chirag wrote:

I have created a small patch to separate DIFlags for DIBasicType. https://reviews.llvm.org/D74470
I will modify this patch afterwards.

Thanks, I will look at that soon.

The use case of this patch is to generate optional dwarf attributes DW_AT_digit_count/DW_AT_decimal_scale/DW_AT_binary_scale/DW_AT_decimal_sign/DW_AT_picture_string for languages like cobol, PL1, basic data types with encoding such as DW_ATE_edited DW_ATE_numeric_string etc.

That's more a statement of the functionality of the patch than the use-case. You have a COBOL or PL/1 or BASIC compiler using LLVM?

• Bitcode backward compatibility, added flag with distinct to mark BTFlags' presence.

In D74470#1871645, @djtodoro wrote:

Does this support the old metadata? We should have a test for the old metadata as well (e.g. flags: DIFlagBigEndian).

As we discussed on D73261, we might have to move a few flags from DIFlags into DISPFlags.

• Updated bitcode test to check type with proper variable.

I have created a small patch to separate DIFlags for DIBasicType. https://reviews.llvm.org/D74470
I will modify this patch afterwards.

Ping.

Fixed as per comments. passed proper execution/register context. for the time being DW_OP_piece is not allowed in call op. Working on testcase.

Missed, LLVMGlobalSetMetadata.

Can you please push this changes, i do not have write permission. Thanks.

Ping.

• clang-format

• Added DW_AT_binary_scale attribute.
• Fixed llvm type code formatting
• Fixed minor typos.

In D52295#1248346, @probinson wrote:

What I meant by "use case" is, do you have a front end that will make use of these new features?

Are you supporting binary scale as well as decimal scale, for [un]signed_fixed encodings? It looks like you are supporting only decimal scale.

@aprantl yes, at the moment clang is not generating bi-endian code, hence not using DIFlags.

• Updated metatdata unittest
• Fixed optional issues with decimal attributes.
• Added doxygen comments.
• Updated asm roundtrip tests.

I will also include the metadata unit test with listed changes, thanks.

Can anyone kindly submit this? i do not have write access. Thanks.

Hello Kristina,

• fixed mistake with go.

Hello whitequark.

• minor format fixes.

This change will support to add optional dwarf attributes to tag_base_type entries.
At the moment only DW_AT_endianity is supported. which is useful for bi-endian code generation such as scaler-storage-order extension in gcc

Hello adrian,

• removed wrong testcase.

• fixed unique node lookup for basicType
• fixed assert message for conflicting flags.
• added testcases for comparing basictype with flags.

• Minor changes relating to DIFlags checks.
• Added bitcode upgrade test for DIBasicType with flags endianity.

The last thing that is missing is a bitcode upgrade test (unless you can show that one of the existing checked in .bc files in the testsuite also covers the upgrade path for the old, shorter DIBasicType).

It seems like there is no such dedicated test case for DIBasicType, but some other testcases such as DIVariable, DiExpression uses the DIBasicType.
Please advice, if you want me to create a new test case for bitcode upgrade of DIBasicType or is it okey to modify the existing DiVariable one.

• moved endianity enums to dwarf.def
• Added verifier check for DIBasicType for invalid flags.
• removed hashing of flags from DIBasicType

• Included Full context.
• Changed flag values for BigEndian, LittleEndian.
• Included assembler round-trip test for BasicType with BigEndian/LittleEndian testcases.

for the time being the basicType that maps to DW_TAG_base_type, but in future the variables created using that basictype or any other type with DW_TAG_variable and DW_TAG_formal_parameters tags.

Do you need to support endinaness attributes just on basic types or eventually on any type?

for the time being the basicType that maps to DW_TAG_base_type, but in future the variables created using that basictype or any other type with DW_TAG_variable and DW_TAG_formal_parameters tags.

For testing i have ran llvm-check with no regressions.

Extended DIFlags to support DW_AT_endianity in basictype as per DWARFv4

Hello Adrian,

Hello Adrian,
can i discard this review?

Testing Done: Compilation and llvm-check.

Thank you guys for the help.

Hello All,