On Mon, Dec 9, 2024 at 9:12 PM Mark Brown <broonie@xxxxxxxxxx> wrote: > > On Mon, Dec 09, 2024 at 08:51:28PM +0800, Weizhao Ouyang wrote: > > On Mon, Dec 9, 2024 at 8:36 PM Mark Brown <broonie@xxxxxxxxxx> wrote: > > > > > // Set SVCR if we're doing SME > > > > - cbz x1, 1f > > > > adrp x2, svcr_in > > > > ldr x2, [x2, :lo12:svcr_in] > > > > + cbz x1, 1f > > > > msr S3_3_C4_C2_2, x2 > > > > This is against an older verison of the code so wouldn't apply now. > > > It's not also checking the value of SVCR, this is checking the SME flag > > > the check is against x1. > > > This patch aims to fix the second check (SVCR_ZA_SHIFT) instead of > > the first one (the x1 SME flag you're referring to): > > If we don't have SME we should be skipping over all the SME code and > never even looking at the value of SVCR. Looking at the current version > of the code it does that, it branches to check_sve_in if SME is not > enabled. Hi Mark, Yes we should skip it, this is just a minor tweak based on the current implementation, after all, we manually passed its value by svcr_in. Which latest code version are you referring to? I think check_sve_in is in fp testcase, not in the abi testcase. (checked the -next tree) BR, Weizhao
