Hi Shimoda-san,
On Thu, Apr 2, 2020 at 5:58 AM Yoshihiro Shimoda
<yoshihiro.shimoda.uh@xxxxxxxxxxx> wrote:
>
> Hi Prabhakar-san,
>
> > From: Yoshihiro Shimoda, Sent: Thursday, April 2, 2020 10:23 AM
> <snip>
> > By the way, according to previous your report, you are using pci/next branch
> > and the branch is based on v5.6-rc1. There is no evidence though,
> > I'd like to use next-20200401 tag from linux-next repo to use v5.6 based kernel
> > whether the strange issue happens on the latest v5.6 kernel code or not.
> > Note that I confirmed the next-20200401 tag has a commit d293237739414d
> > ("misc: pci_endpoint_test: Use streaming DMA APIs for buffer allocation").
>
> I'm afraid but I'd like to recall this because next-20200401 seems unstable on
> R-Car Gen3 environment (a lot of WARING happens). So, perhaps using linux mainline
> v5.6 + merging pci/next branch is better.
>
Taking your advice I rebased my patches on Geert's tree [1] (master
branch) and choose
M3N as my root device, and with rigorous testing I can confirm
read/write/copy work
without any issues. Using the same branch with G2N as root device still failed.
Since this is a Root device issue and not ep I am posting v6 and
swiotlb can be handled
separately for G2x platforms.
[1] https://git.kernel.org/pub/scm/linux/kernel/git/geert/renesas-drivers.git/
Cheers,
--Prabhakar
> Best regards,
> Yoshihiro Shimoda
>
> > Best regards,
> > Yoshihiro Shimoda
>
