• clang-format

• Convert memory region to early return style to make the logic clearer
• Only call ::GetLastError once when VirtualQueryEx fails

This is probably pretty trivial but my first dive into win32 stuff so wanted to double check my logic before relanding.

Links to save combing the document:
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vqshlh_u16
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vqshls_u32
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vqshld_u64
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vqrshlh_u16
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vqrshls_u32
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vqrshld_u64
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vqshlb_u8
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vqrshlb_u8
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vshld_u64
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vrshld_u64

Links instead of searching the document:
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vceqd_s64
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vceqzd_s64
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vcged_s64
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vcgezd_s64
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vcgtd_s64
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vcgtzd_s64
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vcled_s64
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vclezd_s64
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vcltd_s64
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vcltzd_s64
https://developer.arm.com/architectures/instruction-sets/simd-isas/neon/intrinsics?search=vtstd_s64

Rebased

Updated to fix all the vqmovun that have incorrect types.

• Check complete error message instead of using regex

• Return early
• Only check for expected error in TestMemoryRegion

In D87442#2273064, @omjavaid wrote:

This seems fine to me with some minor nits. Also do you plan on writing a Linux API test for this which test memory regions on Linux? I couldnt locate one already written.

• clang-format Minidump file
• Remove HasFlags from API, return True/False from GetFlags instead
• Rename MapKind to MapsKind and make it an enum class

This LGTM but I don't know the API standards very well. Some else should give a final review.

What isn't covered in testing is checking that "memory region" will not print "flags:" if there was no flag info. This would need a way to make a fake lldbserver that didn't reply with the flags field, I could write tests that expect the "flags:" but then when you ran them on a different kernel they might fail.

This is in support of AArch64 memory tagging support as described here: http://lists.llvm.org/pipermail/lldb-dev/2020-August/016402.html
See "### Extending qMemoryRegion"

Thanks for the review!

• Use settings command so we don't have to substr help's output
• Move list and dedent calls into assert function

Combined two functions into one single run command
and check message function.

I realise the test is a bit meta so if there might be a better place to put it. It does at least run successfully from the API test dir.

https://www.llvm.org/docs/GettingStarted.html gives the minimum gcc as 5.1. (I found this cross compiling, the apt package for AArch64 is 5.4.0)

Ok to land this as is just to remove some confusion?

Thanks for the pointer, I found more mistakes this way, e.g.: https://godbolt.org/z/f6n95z
(I don't think I can use <type_traits> in a test but there is clang/test/Headers/x86-64-apple-macosx-types.cpp which uses is_same with it's own implementation so I can use/copy that)

The logic LGTM, I wondered at the time I changed this if it was possible to not have a size but didn't know how to trigger it.

I was surprised too. Perhaps there is a way to write the test file such that any type mismatch is a warning and compile with -Werror (maybe a new file, the IR might get messy). I will spend some time looking into this.

Originally added in https://reviews.llvm.org/D77793.

LGTM, I'll let @chill give the final word.

• Use more standard 'test -f' command before 'mv's

• mv with stderr redirected instead of using non standard file -E flag

• Updated rm/mv command lines to account for $DEFAULT_TRIPLE being the same as %target_triple.

For the record my logic here is that I could change the tool names to use the same name as --version. However that will break whatever mips toolchain needed this code in the first place. (https://reviews.llvm.org/D13340?id=36227#inline-108606)

• Write to cmake and clang's default triple to also fix failure reported by @stevewan.

Right, I see the issue.

• Removed \ from regex, not needed for POSIX syntax.
• Reflowed text in commit description.

I saw similar behaviour on Mac OSX and fixed that in the reland. (https://reviews.llvm.org/rGd6efc9811646edbfe13f06c2676fb469f1c155b1)

• Added check for alternate gcc names in the first test.

LGTM

Relanded as d6efc9811646edbfe13f06c2676fb469f1c155b1. Change was to get the default triple from clang itself, as on MacOS this can be different to the value you set via CMake.