On Tue, Sep 16, 2025 at 1:07 AM Stefan Hajnoczi <[email protected]> wrote:
>
> Hi Alistair,
> I saw you requested access to the opensbi mirror repo on QEMU's GitLab
> project.
>
> I went ahead and pushed the latest OpenSBI master branch to QEMU's mirror:
> https://gitlab.com/qemu-project/opensbi/-/commits/master?ref_type=HEADS
Awesome! Thanks for that
>
> The mirror got stuck because commit a2807646a858 ("include: Adjust
> Sscofpmf mhpmevent mask for upper 6 bits") is not present in the
> upstream repo. There was probably a force push upstream that replaced
> it with commit df997c6e55fe ("include: Adjust Sscofpmf mhpmevent mask
> for upper 8 bits"). The diff of the QEMU mirror commit and the
> upstream commit is:
>
> diff --git a/include/sbi/riscv_encoding.h b/include/sbi/riscv_encoding.h
> index 050674a..2ed05f2 100644
> --- a/include/sbi/riscv_encoding.h
> +++ b/include/sbi/riscv_encoding.h
> @@ -207,7 +207,7 @@
>
> #endif
>
> -#define MHPMEVENT_SSCOF_MASK _ULL(0xFC00000000000000)
> +#define MHPMEVENT_SSCOF_MASK _ULL(0xFF00000000000000)
>
> #define ENVCFG_STCE (_ULL(1) << 63)
> #define ENVCFG_PBMTE (_ULL(1) << 62)
>
> I looked at the upstream discussion about the reserved bits and I
> guess that the author changed their mind and force pushed with an
> updated mask constant. I mention this because I came to the conclusion
> that there was nothing nefarious about the force push and that it's
> safe to update QEMU's mirror to follow upstream again.
Thanks for digging into that, I agree that it seems like someone just
wanted to correct a bit mask.
Alistair
>
> Please let me know if you have any issues updating OpenSBI in QEMU and
> I can tweak the QEMU mirror repository as needed.
>
> Thanks,
> Stefan
