The disassembly of both ARM and Thumb turned out to be matching but CHECK-INSTR-{ARM,THM} can be added in the future for the lines that diverge.

Update the test to align with ELF_static_thumb_reloc.s test.

We've experienced a similar issue and added a fix in an upcoming patch Implement ELF::R_ARM_MOVT_ABS and R_ARM_MOVW_ABS_NC. I don't have commit access so you may revert it until the new patch arrives and fixes it.

Update jitlink-check condition as an explicit offset similar to the change in D157533

Update author email.

update author mail

Update jitlink-check test to indicate the offset explicitly.

(NFC) Removed leading zeros from immediates in comments

(NFC) Updated enc/decode description comment bits with precise number of 0's

(NFC) Correct error message

Thanks @mpaszkowski. Can you land this patch for me?
Please use “Eymen Ünay eymenunay@outlook.com" to commit the change.

Updated test and squashed commits.

In D157533#4641795, @sgraenitz wrote:

Can we align the test with the (now) existing ELF_static_arm_reloc.s please? The rest should be ready to land I think.

Updated test regarding ELF_static_thumb_reloc.s and squashed commits

Closed by commit rG9c017a99d570: [jitlink][rtdyld][checker] Re-apply 4b17c81d5a5 with fixes..

Hi @TWeaver, thanks for reverting. I don't have commit access yet so I appreciate it.

Revision https://reviews.llvm.org/D158280 reopened instead.

Fixed buildbot failures by adding missing plumbing for Subtarget Features and CPU string

New commit description

TargetInfo passing is made safer by passing struct. TT and TF are converted to value members rather than references. Minimised boilerplate.

included both of the unconditional checks

Removed unconditional check and combined commits into one.

Added getJITSymbolFlags to COFF Thumb RuntimeDyld backend to emit Thumb TargetFlag. Fixed a bug in RuntimeDyld/ARM/COFF_Thumb.s.
Added checks to triple switching arm<->thumb into getTripleFromTargetFlag in RuntimDyldChecker.

Currently RuntimeDyld/ARM/COFF_Thumb.s fails due to not finding TargetFlag set to Thumb Flag in SymInfo. I'm still investigating the issue.

• removed hasTargetFlags and updated its calls in aarch32 jitlink
• removed MemoryRegionInfo constructor without the TargetFlag field
• removed irrelevant assert

In D158280#4599171, @lhames wrote:

What if you made RuntimeDyldChecker::getInstPrinter and RuntimeDyld::getDisasssembler take a symbol, then check the target flags in each of those methods so that the callees reduce back to, e.g.

auto Dis = Checker.getDisassembler(Symbol); // TT and TF taken from Checker member variables.

if (auto E = Dis.takeError()) {
  errs() << "Error obtaining disassembler: " << toString(std::move(E))
         << "\n";
  return false; // Return false or take appropriate action.
}

• Moved TargetFlag checking into getDisassembler
• Introduced getTripleFromTargetFlag for backend specialization
• Replaced hasTargetFlag with getTargetFlag

reword comment and reformat code

Addressed feedback

NFC

• Add descr to Arm_Call enum.
• Remove alignment related comments.
• Standardize comment marker as '#'

replace "//" with "#" entirely

Triple override is necessary for targets that do not have a clear architecture signature in their object files. Here you can see the issue on Thumb disassembly.

correct title tag

avoided redefinition of TT

In D157533#4574209, @sgraenitz wrote:

Looks great, thanks! I think at some point we should add a .thumb section with .thumb_func functions to our tests, so we get coverage for the code that handles ARM-Thumb transitions. However, I guess we need ARM stubs for it first?

In D157226#4579993, @sgraenitz wrote:

LGTM, thanks! Should I land it on your behalf?

In D157541#4574325, @sgraenitz wrote:

Thanks!

Is the test actually working right now? Since we always use indirection stubs and we only have a Thumb implementation right now, my quick assumption was that the TargetSymbol should always have the ThumbSymbol flag set (and thus always bail out). Maybe I find the time in the next days to look into it myself and give some informed feedback.

Thanks for pointing out.

Removed non-data relocation checks and standardised comment styling

In D157519#4574123, @sgraenitz wrote:

This is looking very good already! Nice idea to check ARM and Thumb in one test. I am not sure, however, if it's worth the effort. They both produce R_ARM_ABS32 and R_ARM_REL32. The only difference is the alignment and here Thumb alone would be sufficient:

➜ llvm-objdump -r ...
RELOCATION RECORDS FOR [.text]:                   RELOCATION RECORDS FOR [.text]:
OFFSET   TYPE                     VALUE           OFFSET   TYPE                     VALUE
00000004 R_ARM_ABS32              target          00000002 R_ARM_ABS32              target
00000008 R_ARM_REL32              target          00000006 R_ARM_REL32              target
       ^ ARM is on a 4-byte boundary                     ^ Thumb is 2-byte

What we should do at some point (not necessarily in this patch) is to check whether it works on big-endian targets. Add a eb suffix to the arch in order to test it, e.g. thumbebv7-none-linux-gnueabihf

Reverted a -> an changes for acronyms in comments

In D157226#4564085, @sgraenitz wrote:

Let's drop all the an <--> a changes, I think these are fine.

Thanks @fhahn. Can you land this patch for me?
Please use “Eymen Ünay eymenunay@outlook.com" to commit the change.

Thanks @RKSimon. Can you land this patch for me?
Please use “Eymen Ünay eymenunay@outlook.com" to commit the change.

Thanks @DavidSpickett. Can you land this patch for me?
Please use “Eymen Ünay eymenunay@outlook.com" to commit the change.

Thanks @philnik. Can you land this patch for me?
Please use “Eymen Ünay eymenunay@outlook.com" to commit the change.

Thanks @kadircet. Can you land this patch for me?
Please use “Eymen Ünay eymenunay@outlook.com" to commit the change.

Thanks @maksfb. Can you land this patch for me?
Please use “Eymen Ünay eymenunay@outlook.com" to commit the change.

Typo fix on file name in recommended command