[PATCH] Allow encoded 8-bit floating point constants in ARM vmov instructions

Renato Golin renato.golin at linaro.org
Tue Dec 17 14:16:58 PST 2013

On 17 December 2013 20:15, David Peixotto <dpeixott at codeaurora.org> wrote:

> The arm and gcc assmblers both accept the fconstd alias, even in the
> context
> of unified syntax files. While I agree with you that projects should update
> their source code I don't always have the leverage to make that happen :)

Don't trust old assemblers for wisdom, for there is none there. :D

but it looks like that is quite by accident (due to issue 1 above). Do we
> want to support writing these constants or should we only allow
> floating-point literals (e.g. #1.0 instead of #0x70)?

If that's the only problem, I agree this could be fixed. There is no point
in purposely *not* supporting an old feature.

To add to Jim's argument, we (ARM maintainers) have collectively decided
not to actively support pre-UAL syntax unless there was no other way (which
is rare) in UAL.

Pre-UAL, as well as some undocumented GNU extensions could still use the
old assembler while they migrate to a more modern syntax.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20131217/f0b82684/attachment.html>

More information about the llvm-commits mailing list