.if constant expression folding works for .s but not for inline assembly due to UseAssemblerInfoForParsing

[pskocik]

Example:

.text
start:
  nop
end:
.if (end - start) == 0
.err
.endif

compiles when saved as a .s file. But when the same is put into a .c file as:

__asm(
".text\n"
"start:\n"
"  nop\n"
"end:\n"
".if (end - start) == 0\n"
".err\n"
".endif\n"
);

it fails with error: expected absolute expression. GCC succeeds with both examples. #53728 is probably related.

[llvmbot]

@llvm/issue-subscribers-backend-x86

Example: ```s .text start: nop end: .if (end - start) == 0 .err .endif ```

compiles when saved as a .s file. But when the same is put into a .c file as:

__asm(
".text\n"
"start:\n"
"  nop\n"
"end:\n"
".if (end - start) == 0\n"
".err\n"
".endif\n"
);

it fails with error: expected absolute expression. GCC succeeds with both examples. #53728 is probably related.

[MaskRay]

it fails with error: expected absolute expression. GCC succeeds with both examples. #53728 is probably related.

clang -c -fno-integrated-as a.c works as well. I have some notes in https://maskray.me/blog/2023-06-18-port-llvm-xray-to-apple-systems

In the case of direct object emission, where LLVM IR is directly lowered to an object file bypassing assembly (e.g., using clang -c a.c instead of clang -c a.s or clang -c --save-temps a.c), the assembler information is not used at parse time (MCStreamer::UseAssemblerInfoForParsing). As a result, the assembly code .quad (Lxray_sleds_end0-lxray_sleds_start0)>>5 will be transformed into a fixup. Lxray_sleds_end0 is a pending label which will then be assigned to the next fragment when AsmPrinter emits new piece of data to the xray_instr_map section.

---

@niravhdave
Commit 6c0665e
(https://reviews.llvm.org/D45164) enabled certain constant expression
evaluation for MCObjectStreamer at parse time (e.g. .if directives,
see llvm/test/MC/AsmParser/assembler-expressions.s).

getUseAssemblerInfoForParsing was added to make clang -c handling
inline assembly similar to MCAsmStreamer (e.g. llvm-mc -filetype=asm),
where such expression folding (related to
AttemptToFoldSymbolOffsetDifference) is unavailable.

I believe this is overly conservative. We can make some parse-time
expression folding work for clang -c even if clang -S would still
report an error, a MCAsmStreamer issue (we cannot print .if
directives) that should not restrict the functionality of
MCObjectStreamer.

% cat b.cc
asm(R"(
.pushsection .text,"ax"
.globl _start; _start: ret
.if . -_start == 1
  ret
.endif
.popsection
)");
% gcc -S b.cc && gcc -c b.cc
% clang -S -fno-integrated-as b.cc     # succeeded

% clang -c b.cc     # succeeded with this patch
% clang -S b.cc     # still failed
<inline asm>:4:5: error: expected absolute expression
    4 | .if . -_start == 1
      |     ^
1 error generated.

[musamaanjum]

I've hit the same bug. Clang gives an error with the following extended asm:

<inline asm>:7:8: error: expected absolute expression
    7 |         .ifne . - test_page - 4096
      |               ^
<inline asm>:8:2: error: test page is not one page long
    8 |         .error "test page is not one page long"
      |         ^

asm (
	".pushsection \".text\", \"ax\"\n\t"
	".balign 4096\n\t"
	"test_page: .globl test_page\n\t"
	".fill 4094,1,0xcc\n\t"
	"test_syscall_insn:\n\t"
	"syscall\n\t"
	".ifne . - test_page - 4096\n\t"
	".error \"test page is not one page long\"\n\t"
	".endif\n\t"
	".popsection"
    );

But clang builds fine if the asm is moved to the assembly file, or by generating the asm by GCC and then building the asm to binary by clang.

[MaskRay]

asm (
".pushsection ".text", "ax"\n\t"
".balign 4096\n\t"
"test_page: .globl test_page\n\t"
".fill 4094,1,0xcc\n\t"
"test_syscall_insn:\n\t"
"syscall\n\t"
".ifne . - test_page - 4096\n\t"
".error "test page is not one page long"\n\t"
".endif\n\t"
".popsection"
);

#91082 will make this code compile with clang -c but still not with clang -S -fintegrated-as.

I've also have experiments [MC] Enable getAssemblerPtr for MCAsmStreamer to hopefully make MCAsmStreamer fold more expressions similar to MCObjectStreamer, but there is a lot of complexity.

[MaskRay]

Created https://discourse.llvm.org/t/rfc-clang-assembly-object-equivalence-for-files-with-inline-assembly/78841
(also mentioned at #91082 )