[Macchiato] ODP: Fwd: SAS/SATA card issues

[Marcin Wojtas]

Hi Baruch,

czw., 17 sty 2019 o 12:41 Baruch Siach <baruch at tkos.co.il> napisał(a):
>
> Hi Marcin, Andy,
>
> On Thu, Jan 17 2019, Marcin Wojtas wrote:
> > czw., 17 sty 2019 o 11:15 <rattusrattus at debian.org> napisał(a):
> >> kernel is 4.19.0-1-arm64 -  Debian stock as reported by uname
> >> [4.19.0-1-arm64 #1 SMP Debian 4.19.12-1 (2018-12-22) aarch64]
> >> Again am I reporting the correct information?
> >
> > Yes, that's sufficient. Recently Baruch fixed the PCIE reset handling
> > in the Linux kernel on McBin (merged to linux-next a week ago), so the
> > thing you see may be related.
>
> The PCIe reset gpio DT fix[1] is in v5.0-rc2. This fix has no effect
> without the PCIe driver reset handling patch[2] in v5.0-rc1. But the
> combination of both actually breaks PCIe on the Macchiatobin. So I sent
> a revert[3].
>

Thanks for the explanation.

> I tested these patches with U-Boot only. Not sure how EDK handles the
> Macchiatobin PCIe serdes initialization and reset GPIO.
>

EDK2 handles the reset in similar manner as u-boot and later it should
be not executed by OS, afaik.

Serdes is initialized by the same code in BL31 via smc calls, so this
code is common for all components (u-boot, uefi, linux).

After HW fix (board v1.3), executing reset and using common serdes
code, I had no reports about PCIE instability, therefore I am curious,
what is going on with Andy's card.

Best regards,
Marcin

> baruch
>
> [1] https://git.kernel.org/linus/59c4dccbc3676144091783c8b46bd51daa4f80bc
>
> [2] https://git.kernel.org/linus/3d71746c420c1c1c27cf5c4e48f8fa0a6cfdc185
>
> [3] http://patchwork.ozlabs.org/patch/1024067/
>
> --
>      http://baruch.siach.name/blog/                  ~. .~   Tk Open Systems
> =}------------------------------------------------ooO--U--Ooo------------{=
>    - baruch at tkos.co.il - tel: +972.52.368.4656, http://www.tkos.co.il -