Hi Richard,
> I think this should be "lk*r", not "l*rk". SP is only going to crop up
> in rare circumstances, but we are always going to need this pattern if
> it does and hiding this from register preferencing is pointless. It's
> not like the compiler is going to start allocating SP in the
On 24/07/2019 16:16, Wilco Dijkstra wrote:
The Thumb-2 movsi patterns try to prefer low registers for loads and stores.
However this is done incorrectly by using 2 separate variants with 'l' and 'h'
register classes. The register allocator will only use low registers, and
as a result we end u
Hi Wilco,
On 7/24/19 4:16 PM, Wilco Dijkstra wrote:
The Thumb-2 movsi patterns try to prefer low registers for loads and
stores.
However this is done incorrectly by using 2 separate variants with 'l'
and 'h'
register classes. The register allocator will only use low registers, and
as a result
The Thumb-2 movsi patterns try to prefer low registers for loads and stores.
However this is done incorrectly by using 2 separate variants with 'l' and 'h'
register classes. The register allocator will only use low registers, and
as a result we end up with significantly more spills and moves to hi
