This change makes a tpb32l build work for ARM before ARMv6 — including the armel Debian build, as opposed to armhf.
The main obstacle to supporting older Arm processors directly is that the "arm32.ss" backend currently uses hardware-supported floating point instructions via the VFP extension, which means ARMv6 and later. A software floating-point mode could be added (or restored, since it looks like there may have been one at some point?), but my guess is that there isn't any demand at the moment. Meanwhile, this simple change at least lets tpb32l build.
This change makes a
tpb32l
build work for ARM before ARMv6 — including thearmel
Debian build, as opposed toarmhf
.The main obstacle to supporting older Arm processors directly is that the "arm32.ss" backend currently uses hardware-supported floating point instructions via the VFP extension, which means ARMv6 and later. A software floating-point mode could be added (or restored, since it looks like there may have been one at some point?), but my guess is that there isn't any demand at the moment. Meanwhile, this simple change at least lets
tpb32l
build.See also racket/racket#4955.