Re: [Pixman] [PATCH] Add support for aarch64 neon optimization

2016-04-01 Thread Siarhei Siamashka
ad of PLD in the beginning of each new scanline, because Cortex-A8 ignored PLD prefetches on TLB misses. Well, nowadays modern ARM processors got automatic hardware prefetchers, which greatly reduce or eliminate the need for doing prefetch in software. Here is a test patch for pixman, which is di

[Pixman] [PATCH] ARM: NEON: Use AArch64 compatible syntax for barrel shifter arguments

2016-04-01 Thread Siarhei Siamashka
The following 32-bit ARM code mov TMP1, X, asr #16 add TMP1, TOP, TMP1, asl #2 needs to be converted into asr TMP1, X, #16 add TMP1, TOP, TMP1, lsl #2 in order to be accepted by both 32-bit and 64-bit ARM assemblers. The conversion has been done using the