On Tue, May 30, 2023 at 2:38 AM Geert Uytterhoeven <geert@xxxxxxxxxxxxxx> wrote: > > Hi Adam, > > On Fri, May 26, 2023 at 9:59 PM Adam Ford <aford173@xxxxxxxxx> wrote: > > On Fri, May 26, 2023 at 2:13 AM Geert Uytterhoeven <geert@xxxxxxxxxxxxxx> wrote: > > > On Fri, May 26, 2023 at 6:07 AM Adam Ford <aford173@xxxxxxxxx> wrote: > > > > On Mon, May 22, 2023 at 3:06 AM Geert Uytterhoeven <geert@xxxxxxxxxxxxxx> wrote: > > > > > On Sat, May 20, 2023 at 5:13 AM Adam Ford <aford173@xxxxxxxxx> wrote: > > > > > > I am trying to see if the Open Source PowerVR driver [1] for the 6250 > > > > > > can be made to work on the RZ/G2M, and I am having difficulty getting > > > > > > the power domain to turn on. > > > > > > > > > > > > In the GPU node, I set: > > > > > > power-domains = <&sysc R8A774A1_PD_3DG_B>; > > > > > > > > > > LGTM... > > > > > > > > > > > Unfortunately, when it boots, I get the following error message: > > > > > > > > > > > > [ 8.313305] powervr fd000000.gpu: error -ENOENT: failed to add to > > > > > > PM domain 3dg-b > > > > > > > > > > That means genpd_add_device() failed (but why?)... > > > > > > > > With a pointer from Marek V, I appear to have the necessary clock and > > > > the power domain operational by borrowing from the down-stream RZ/G2 > > > > kernel. Even if I cannot get the PVR driver working with this kernel, > > > > would you accept the clock and power domain patch now, or do you want > > > > me to wait until we have a user of these (aka PVR) working? > > > > > > I don't think it makes much sense to add the clock and power domain > > > patches upstream now, as they impact system behavior, while we don't > > > know if they are sufficient to make the PVR work. > > > The clock patches are probably (more or less) OK, and they do not have > > > much impact when the PVR is not enabled or not used. > > > For the power domain patches, it's different: some of them are quite > > > intrusive, and we still don't know the full rationale behind them. > > > Out of curiosity, which power domain patch(es) did you have to apply? > > > > I didn't really have to change the power domains per se, but I had to > > add the clocks that the GPU power domains needed and I refereneced the > > power-domain from the GPU node. > > OK. > > > Do you want me to send the patch as an RFC? > > I guess you mean patch series[1], submitted by Marek? > That one is marked "Chances Requested" in patchwork[2]. Moving it > forward means fixing the issues pointed out. I edited different files based on the downstream kernel from Renesas. It also modifies different files, but in a similar way. I was doing this on an r8a774a1 (RZ/G2M). I tried to address the concerns you sent to Marek. I'll post the RFC momentarily. It's a 3-part update to add a generic Rcar-gen3 clock, add the 3dge clock, then the add GPU node. adam > > Thanks! > > [1] https://lore.kernel.org/all/20220314220012.218731-1-marek.vasut@xxxxxxxxx/ > [2] https://patchwork.kernel.org/project/linux-renesas-soc/list/?series=623329&state=* > > Gr{oetje,eeting}s, > > Geert > > -- > Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx > > In personal conversations with technical people, I call myself a hacker. But > when I'm talking to journalists I just say "programmer" or something like that. > -- Linus Torvalds
