riscv failures

[Marco Elver]

On Fri, 6 Jun 2025 at 13:43, Alexandre Ghiti <alex at ghiti.fr> wrote:
>
> Hi Marco,
>
> On 6/5/25 10:49, Marco Elver wrote:
> > On Wed, 4 Jun 2025 at 13:03, Alexandre Ghiti <alex at ghiti.fr> wrote:
> >> Hi Alexander,
> >>
> >> On 6/4/25 10:35, Alexander Potapenko wrote:
> >>> Hi Alexandre,
> >>>
> >>> We switched the ci-qemu2-riscv64 instance to
> >>> "git://git.kernel.org/pub/scm/linux/kernel/git/riscv/linux.git
> >>> for-next", because the "fixes" branch was quite stale and didn't
> >>> contain the latest fixes.
> >>>
> >>> However right now syzkaller is failing to SSH into the instance
> >>> because of some boot-time problems (see below).
> >>> Do you know what could be causing this?
> >>
> >> Yes, we need both commit dd133162c9cf ("ACPI: platform_profile: Avoid
> >> initializing on non-ACPI platforms") merged in 6.16 and commit
> >> 07c9214c790e ("mm/cma: make detection of highmem_start more robust")
> >> merged in late 6.15.
> >>
> >> Can you consider using my fixes branch instead? It will be updated more
> >> frequently. My branch is
> >> https://git.kernel.org/pub/scm/linux/kernel/git/alexghiti/linux.git/log/?h=alex-fixes.
> > We can only pick one for riscv, and can't keep switching the tree.
> > The canonical subsystem tree per MAINTAINERS is:
> > git.kernel.org/pub/scm/linux/kernel/git/riscv/linux.git
> > We thought that for-next should not be continuously broken because it
> > also receives -next exposure.
>
>
> Sorry about this situation, unfortunately it's out of my control,
> hopefully that will change some time soon, but for now I'm only a "R:"
> in MAINTAINERS...

What's the defacto state of the main RISCV tree? Is it broken,
unmaintained, abandoned?

> > So the main question would be: what is the most useful tree to fuzz
> > for riscv? If you declare that your fixes branch is the latest
> > up-to-date riscv branch that will be maintained and updated
> > indefinitely, we can switch (however, it would have been good to see
> > such a branch in the riscv/linux.git repo). Either way, the choice is
> > yours.
>
>
> I can guarantee that my branch will be uptodate, no worries about that
> so if you can switch, it would be nice.
>
> Sorry about for this mess and thanks,

Understood - but I'd like to understand why it's not possible to
update for-next (which would also benefit -next) with the necessary
patches that would unbreak syzbot RISCV fuzzing. This is the normal
process - is the process broken, or are the patches still under
review?