This is the mail archive of the
mailing list for the GCC project.
Re: Potential bug with wide_int_storage::set_len
- From: Eric Botcazou <ebotcazou at adacore dot com>
- To: "Andre Vieira (lists)" <Andre dot SimoesDiasVieira at arm dot com>
- Cc: gcc at gcc dot gnu dot org
- Date: Thu, 13 Oct 2016 12:16:03 +0200
- Subject: Re: Potential bug with wide_int_storage::set_len
- Authentication-results: sourceware.org; auth=none
- References: <57FE5EA6.email@example.com> <3682121.kGSTZQmubs@polaris> <57FF49F0.firstname.lastname@example.org>
> If set_len really does 'ignore' signedness, then we are not treating X
> as having signedness SGN.
It does not, my interpretation is that it sign-extends when the value is not
already sign-extended, probably for the reason exposed by Richard. But that's
admittedly quite confusing so a comment by the author would be in order.