Hi Palmer, On Fri, Sep 6, 2019 at 3:00 AM Palmer Dabbelt <pal...@sifive.com> wrote: > > On Thu, 05 Sep 2019 08:25:46 PDT (-0700), bmeng...@gmail.com wrote: > > Hi Alistair, > > > > On Thu, Sep 5, 2019 at 3:50 AM Alistair Francis <alistai...@gmail.com> > > wrote: > >> > >> On Sat, Aug 31, 2019 at 7:54 PM Bin Meng <bmeng...@gmail.com> wrote: > >> > > >> > As of today, the QEMU 'sifive_u' machine is a special target that does > >> > not boot the upstream OpenSBI/U-Boot firmware images built for the real > >> > SiFive HiFive Unleashed board. Hence OpenSBI supports a special platform > >> > "qemu/sifive_u". For U-Boot, the sifive_fu540_defconfig is referenced > >> > in the OpenSBI doc as its payload, but that does not boot at all due > >> > to various issues in current QEMU 'sifive_u' machine codes. > >> > > >> > This series aims to improve the emulation fidelity of sifive_u machine, > >> > so that the upstream OpenSBI, U-Boot and kernel images built for the > >> > SiFive HiFive Unleashed board can be used out of the box without any > >> > special hack. > >> > > >> > The major changes include: > >> > - Heterogeneous harts creation supported, so that we can create a CPU > >> > that exactly mirrors the real hardware: 1 E51 + 4 U54. > >> > - Implemented a PRCI model for FU540 > >> > - Implemented an OTP model for FU540, primarily used for storing serial > >> > number of the board > >> > - Fixed GEM support that was seriously broken on sifive_u > >> > - Synced device tree with upstream Linux kernel on sifive_u > >> > > >> > OpenSBI v0.4 image built for sifive/fu540 is included as the default > >> > bios image for 'sifive_u' machine. > >> > > >> > The series is tested against OpenSBI v0.4 image for sifive/fu540 > >> > paltform, U-Boot v2019.10-rc1 image for sifive_fu540_defconfig, > >> > and Linux kernel v5.3-rc3 image with the following patch: > >> > > >> > macb: Update compatibility string for SiFive FU540-C000 [1] > >> > > >> > OpenSBI + U-Boot, ping/tftpboot with U-Boot MACB driver works well. > >> > Boot Linux 64-bit defconfig image, verified that system console on > >> > the serial 0 and ping host work pretty well. > >> > > >> > An OpenSBI patch [2] was sent to drop the special "qemu/sifive_u" > >> > platform > >> > support in OpenSBI. The original plan was to get the drop patch applied > >> > after this QEMU series is merged. However after discussion in the OpenSBI > >> > mailing list, it seems the best option for us is to let OpenSBI continue > >> > shipping the special "qemu/sifive_u" platform support to work with QEMU > >> > version <= 4.1 and deprecate the support sometime in the future. A patch > >> > will need to be sent to OpenSBI mailing list to update its document. > >> > > >> > v4 is now rebased on Palmer's QEMU RISC-V repo "for-master" branch. > >> > Dropped the following v3 patch that was already done by someone else. > >> > - riscv: sifive_u: Generate an aliases node in the device tree > >> > - riscv: sifive_u: Support loading initramfs > >> > >> I'm having trouble applying this. Do you mind sharing a public git branch? > > > > So I see Palmer has rebased his "for-master" branch against QEMU > > master and now this patch series cannot be applied cleanly. > > > > Even worse, the current "for-master" branch has build errors for QEMU > > RISC-V. > > Sorry, I haven't gotten around to fixing up the patch queue for now. It's > always OK to just send stuff against Peter's master, particularly if it's a > big > patch set like this, as that's more likely to be a stable base for testing.
I was originally using Peter's branch but later I got to know there was some patch conflicts between my patch series and what was already applied in the RISC-V queue. Hence I switched to rebase my series on "for-master" since v4. I will rebase my patch series again, and send out a v8 (hopefully the last version) soon. Regards, Bin
