https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92921
Bug ID: 92921
Summary: Feature request: Automatic transliteration of AT&T
inline asm into Intel syntax
Product: gcc
Version: 9.2.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: inline-asm
Assignee: unassigned at gcc dot gnu.org
Reporter: warp at iki dot fi
Target Milestone: ---
gcc defaults to using AT&T asm syntax for x86 targets, with Intel syntax being
an option (with -masm=intel). As a consequence, any inline asm expression is
expected to likewise use AT&T syntax by default, and provide the Intel syntax
as an alternative (using the "{...|...}" syntax). If only the AT&T syntax is
provided in the inline asm expression, then gcc will just blindly copy it as-is
into the output even if the output is in Intel syntax, thus resulting in
incorrect asm.
clang, however, is able to transliterate inline asm written in AT&T syntax into
Intel syntax, when -masm=intel is specified, and thus the inline asm expression
doesn't need to provide both.
It would be nice if gcc had this support as well, as it makes writing inline
asm significantly more convenient (especially when large amounts of it need to
be written).
(It would also be a nice-to-have feature if one would be able to specify in the
asm() statement which syntax it's using, so one could write in Intel syntax
even when the compiler is outputting in AT&T syntax.)
