On 11/06/13 09:45, James Greenhalgh wrote:
Hi,
This patch is a respin of the aarch-common improvements to use a generic
search to find SETs and the variety of shifts, rather than relying on the
hope that we will find something well formed.
I've bootstrapped the patch on a chromebook with -mcpu=cortex-a7,
-mcpu=cortex-a9 and -mcpu=cortex-a15 with no issues, and run
the aarch64-none-elf regression suite with no issues. And I've thrown
a bunch of smoke tests at this over all supported -mcpu values
with no issues, and only a few minor scheduling differences where the
old code was too liberal and did the wrong thing.
Is this OK?
This is OK - thanks,
regards
Ramana
Thanks,
James
---
gcc/
2013-11-06 James Greenhalgh <james.greenha...@arm.com>
* config/arm/aarch-common.c
(search_term): New typedef.
(shift_rtx_costs): New array.
(arm_rtx_shift_left_p): New.
(arm_find_sub_rtx_with_search_term): Likewise.
(arm_find_sub_rtx_with_code): Likewise.
(arm_early_load_addr_dep): Add sanity checking.
(arm_no_early_alu_shift_dep): Likewise.
(arm_no_early_alu_shift_value_dep): Likewise.
(arm_no_early_mul_dep): Likewise.
(arm_no_early_store_addr_dep): Likewise.
