On Mon, 2019-07-01 at 19:01 +0100, Peter Maydell wrote: > On Mon, 1 Jul 2019 at 18:50, Alistair Francis < > alistair.fran...@wdc.com> wrote: > > On Mon, 2019-07-01 at 17:54 +0100, Peter Maydell wrote: > > > On Thu, 27 Jun 2019 at 16:24, Palmer Dabbelt <pal...@sifive.com> > > > wrote: > > > > From: Alistair Francis <alistair.fran...@wdc.com> > > > > > > > > Add OpenSBI version 0.3 as a git submodule and as a prebult > > > > binary. > > > > > > > > Signed-off-by: Alistair Francis <alistair.fran...@wdc.com> > > > > Reviewed-by: Bin Meng <bmeng...@gmail.com> > > > > Tested-by: Bin Meng <bmeng...@gmail.com> > > > > Signed-off-by: Palmer Dabbelt <pal...@sifive.com> > > > > --- > > > > .gitmodules | 3 ++ > > > > Makefile | 5 +- > > > > pc-bios/opensbi-riscv32-virt-fw_jump.bin | Bin 0 -> 28848 > > > > bytes > > > > pc-bios/opensbi-riscv64-sifive_u-fw_jump.bin | Bin 0 -> 28904 > > > > bytes > > > > pc-bios/opensbi-riscv64-virt-fw_jump.bin | Bin 0 -> 28904 > > > > bytes > > > > roms/Makefile | 48 > > > > ++++++++++++++- > > > > ---- > > > > roms/opensbi | 1 + > > > > 7 files changed, 44 insertions(+), 13 deletions(-) > > > > create mode 100644 pc-bios/opensbi-riscv32-virt-fw_jump.bin > > > > create mode 100644 pc-bios/opensbi-riscv64-sifive_u- > > > > fw_jump.bin > > > > create mode 100644 pc-bios/opensbi-riscv64-virt-fw_jump.bin > > > > create mode 160000 roms/opensbi > > > > > > > > diff --git a/.gitmodules b/.gitmodules > > > > index 2857eec76377..7a10e72e09cd 100644 > > > > --- a/.gitmodules > > > > +++ b/.gitmodules > > > > @@ -55,3 +55,6 @@ > > > > [submodule "slirp"] > > > > path = slirp > > > > url = https://git.qemu.org/git/libslirp.git > > > > +[submodule "roms/opensbi"] > > > > + path = roms/opensbi > > > > + url = https://github.com/riscv/opensbi.git > > > > > > It would be nice to state the license for new blobs we're > > > adding to the source tree... In particular, I find the > > > readme at https://github.com/riscv/opensbi a bit confusing > > > because it says it's 2-BSD but also that some parts are Apache- > > > 2.0; > > > the latter is *not* GPL-2.0 compatible. That makes it not > > > completely > > > obvious to me that we can ship this with QEMU. > > > > Ah, I didn't realise there is Apache-2.0 code in there. > > > > In this case all of the code we are using (we aren't using the > > Kendryte > > code with QEMU) is BSD-2-Clause. Even then though does it matter if > > it > > is GPL compatible? This isn't being linked with QEMU it is just > > being > > included with QEMU. > > Yeah, if we're not building the apache-2.0 parts then I think > we're OK, and as you say there's the question of whether QEMU's > GPL affects what we ship as mere bios blobs to run as guest > code anyway. But it's sufficiently not a "really obviously ok" > that I'd like a second opinion; cc'd some people who might have > second opinions. > > > I can add details of the license in the source tree, where should > > it be > > added? > > I was thinking in the commit message. We also seem to > have human-readable information in pc-bios/README for > the various blobs, which seems to be a good place for a brief > one-para summary of what OpenSBI is, a URL for its upstream, > and a brief note of the license.
Ok, I'll add that. > > > > Also, new git modules in .gitmodules should be qemu.org > > > URLs, not random external ones. (ie, we should set up > > > mirroring of any new external repo we start shipping > > > code and binaries for). We can set this up and fix up the > > > gitmodules file after the fact, but the ideal is to do > > > it in advance rather than afterwards. > > > > Ok, who can setup a git mirror for the QEMU repo? > > That would be Jeff or Stefan (cc'd). > > > PS: It seems like there are still some issues with this patch so > > maybe > > it's worth dropping this patch from the PR so that we can get > > everything else merged. > > Yeah, that would be my suggestion. Can you do that when you merge it or does Palmer have to send the PR again? Alistair > > thanks > -- PMM
