|
From: | Richard Henderson |
Subject: | Re: [PATCH v3] target/arm: Fix sve2 ldnt1 and stnt1 |
Date: | Tue, 8 Mar 2022 07:12:41 -1000 |
User-agent: | Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101 Thunderbird/91.5.0 |
On 3/8/22 01:56, Peter Maydell wrote:
-# SVE2 64-bit gather non-temporal load -# (scalar plus unpacked 32-bit unscaled offsets) +# SVE2 64-bit gather non-temporal load (scalar plus 64-bit unscaled offsets) LDNT1_zprz 1100010 msz:2 00 rm:5 1 u:1 0 pg:3 rn:5 rd:5 \ - &rprr_gather_load xs=0 esz=3 scale=0 ff=0 + &rprr_gather_load xs=2 esz=3 scale=0 ff=0We correct the xs= value here...
...
+ switch (a->esz) { + case MO_32: + fn = gather_load_fn32[mte][be][0][0][a->u][a->msz]; + break; + case MO_64: + fn = gather_load_fn64[mte][be][0][2][a->u][a->msz]; + break; + }...but then instead of using it we hard code use of 0 or 2 here, which makes the change above a bit moot.
Yeah, sorry for that. I was in the middle of reducing the argument set collected by those decode patterns, then decided there was little advantage, and left something in the middle.
r~
[Prev in Thread] | Current Thread | [Next in Thread] |