Iain Sandoe <i...@sandoe.co.uk> writes: > Hi Richard, > > Richard Sandiford <richard.sandif...@arm.com> wrote: > >> Iain Sandoe <i...@sandoe.co.uk> writes: > >>> The armv8_arm manual [C6.2.226, ROR (immediate)] uses a # in front >>> of the immediate rotation quantity. >>> >>> Although, it seems, GAS is able to infer the # (or is leninent about >>> its absence) assemblers based on the LLVM back end expect it and error >>> out. >>> >>> tested on aarch64-linux-gnu (gcc115) and aarch64-darwin20 (experimental) >> >> Sorry for the slow reply, didn't see this till now. > > Hmm .. I did CC you directly ( but having some troubles with this ISP which > I am > trying to resolve - emails going missing or have to be re-sent … :/ )
Yeah, I got the message, I just didn't see it, sorry. >> The patch is OK in principle, and personally I prefer “#”. But how far >> does this spread? Are only ROR modifiers on logical patterns affected? >> Or is the use of a paranthesised expression instead of a literal the thing >> that makes the difference? > > perhaps, Trying it out locally, that does seem to be the difference: and x1, x2, x3, ror 1 // OK and x1, x2, x3, ror (1) // Rejected and x1, x2, x3, ror #(1) // OK Same for the other modifiers I tried. That doesn't look intentional, but whether it's intentional obviously isn't important in this context. So yeah, the patch is OK. Thanks, Richard