On 2022/12/05 6:14, marius@xxxxxxxxxxxxxx wrote: > December 4, 2022 12:23 PM, "Pali Rohár" <pali@xxxxxxxxxx> wrote: > >>> What next? Should I contact David Milburn <dmilburn@xxxxxxxxxx> (the >>> commit author) and ask for a better fix? >> >> That is a good idea. > > I think I'll wait for an opinion from Damien. Since the same libata-core.c > works perfectly fine with sata-mv driver, but not ahci-mvebu, the problem > might still be in the ahci-mvebu. The comment said: "sata_down_spd_limit > should return if driver has not recorded sstatus speed". This probably means > that ahci-mvebu does not record the expected "sstatus speed". I don't know > what that is, but Damien probably does. Attending OSS Japan conference Monday & Tuesday, so I will be able to look at this second half of this week. Given this latest information, there are a couple of things we can try I think. > >>> @ Pali Rohar: I could't find the errata you mentioned. If you have it, >>> please take a look at the commit above and see if it is affected in any. >>> Thanks. >> >> I do not have more details than what I already wrote in previous email. >> >> Errata documents are in most cases available on Marvell Customer Portal. >> You need to register for accessing them. https://www.marvell.com/portal >> >> Documents which were available publicly without registration are here: >> https://www.kernel.org/doc/html/latest/arm/marvell.html (errata for a38x >> was not publicly available) > > There is a slim chance that I could understand them. I'm just an amateur. I > did some bare-metal programming on MCUs, but the kernel is way too > complicated for me. > -- Damien Le Moal Western Digital Research
