Do you have commit access?

Why not for O32?

powerpc's behavior is definitely correct. AArch64 and MIPS's permissive behaviors are not good. A symbolic value (which can be 32-bit/64-bit) requires more bits than the immediate bits. I think flagging the instruction in the case of a symbolic value is fine.

FWIW: I filed a feature request for GNU as: https://sourceware.org/bugzilla/show_bug.cgi?id=26126

Just workaround as GNU toolchain:

How about?

    .text
main:
    nop
    daddiu $t9, $ra, main

Behaviour the same as GNU toolchain:

In D81908#2095335, @atanasyan wrote:

Does this patch depend on D81919?

0000000120000a70 <main>:                   
   120000a70:   00000000    nop            
   120000a74:   64840000    daddiu  a0,a0,0 # NOTE: still zero :(
    ...

Hi Dan'l,

Hi Alex,

Hi Alex,

Hi Alex,

Hi LLVM developers,

Hi LLVM developers,

Hi Alex,

Hi Chen,

Dear Dylan,

Dear Dylan,

Ping

Nicely ping Dylan :)

Provided a diagnostic indicating that "xpsr" should be used instead of "psr":

Also add an alias for psr_nzcvq about this usecase.

The same encoding as GNU Arm Embedded Toolchain:

Dear Dylan,

Rebase for rL316076 and nicely Ping Dylan and Matt :)

Dear Henry,

Rebase for rL315531

Remove hardcoded macro REFACTORY_PARSEDIRECTIVE.

$ llvm-mc -filetype=obj -triple=avr-unknown-linux -mcpu=atmega328p relax.s -o avr-relax-llvm.o

Refactory EmitValue for AVRMCELFStreamer.

Supported .byte lo8(foo) now :)

Dear Dylan,

Dear Deid,

Dear Dylan,

Waiting for D38304 about Refactoring ParseDirective.

Does it need to Refactory ParseDirective for adding parameter OperandVector Operands to support Modifier?

relocations.s testcase:

...
.short foo
.short gs(foo)
...

Change RELOC_NUMBER for AVR Relocation DIFF Types.

Implement Parse R_AVR_LO8_LDI_NEG and other R_AVR_XXX_NEG Expression for GCC compatible.

Dear Dylan,

Dear Rui,

Add testcase for tinyAVR about R_AVR_LDS_STS_16:

Dear Rui,

Add testcase for R_AVR_LDS_STS_16.

Dear Dylan,

Add ARC Specific e_flags.

Dear Rui,