[PATCH v6 0/2] riscv: Use PUD/P4D/PGD pages for the linear mapping

[Alexandre Ghiti]

On Tue, Mar 7, 2023 at 2:36 PM Alexandre Ghiti <alexghiti at rivosinc.com> wrote:
>
> On Tue, Mar 7, 2023 at 2:19 PM Anup Patel <apatel at ventanamicro.com> wrote:
> >
> > Hi Alex,
> >
> > On Tue, Mar 7, 2023 at 12:22 AM Alexandre Ghiti <alexghiti at rivosinc.com> wrote:
> > >
> > > Hi Anup,
> > >
> > > On Mon, Mar 6, 2023 at 5:33 PM Anup Patel <apatel at ventanamicro.com> wrote:
> > > >
> > > > On Wed, Mar 1, 2023 at 1:56 PM Alexandre Ghiti <alexghiti at rivosinc.com> wrote:
> > > > >
> > > > > This patchset intends to improve tlb utilization by using hugepages for
> > > > > the linear mapping.
> > > > >
> > > > > base-commit-tag: v6.2-rc7
> > > > >
> > > > > v6:
> > > > > - quiet LLVM warning by casting phys_ram_base into an unsigned long
> > > > >
> > > > > v5:
> > > > > - Fix nommu builds by getting rid of riscv_pfn_base in patch 1, thanks
> > > > >   Conor
> > > > > - Add RB from Andrew
> > > > >
> > > > > v4:
> > > > > - Rebase on top of v6.2-rc3, as noted by Conor
> > > > > - Add Acked-by Rob
> > > > >
> > > > > v3:
> > > > > - Change the comment about initrd_start VA conversion so that it fits
> > > > >   ARM64 and RISCV64 (and others in the future if needed), as suggested
> > > > >   by Rob
> > > > >
> > > > > v2:
> > > > > - Add a comment on why RISCV64 does not need to set initrd_start/end that
> > > > >   early in the boot process, as asked by Rob
> > > > >
> > > > > Alexandre Ghiti (2):
> > > > >   riscv: Get rid of riscv_pfn_base variable
> > > > >   riscv: Use PUD/P4D/PGD pages for the linear mapping
> > > >
> > > > I tried this series but it is getting stuck after reaching user space.
> > > >
> > > > Does this series require some other dependent patches ?
> > >
> > > No it should not. Let me take a look: what's your config and the base commit?
> >
> > Please try the alexghiti_test branch at:
> > https://github.com/avpatel/linux.git
> >
> > Compile the kernel with defconfig and launch QEMU as follows:
> > qemu-system-riscv64 -M virt -m 1G -nographic -bios
> > opensbi/build/platform/generic/firmware/fw_jump.bin -kernel
> > ./build-riscv64/arch/riscv/boot/Image -append "root=/dev/ram rw
> > console=ttyS0 earlycon" -initrd ./rootfs_riscv64.img
> >
> > In the above command, rootfs_riscv64.img is a busybox based rootfs.
>
> Ok I can reproduce the problem. I'll debug that a bit further but
> increasing memory allows booting to userspace.
>

This is a real bug and that's because I should split the 1G page that
contains the kernel text alias (and anything that we map with stricter
rights) or something similar. I'll get back soon with a fix.

Thanks Anup for the report,

Alex

> >
> > Regards,
> > Anup
> >
> > >
> > > >
> > > > Regards,
> > > > Anup
> > > >
> > > > >
> > > > >  arch/riscv/include/asm/page.h | 19 +++++++++++++++++--
> > > > >  arch/riscv/mm/init.c          | 28 ++++++++++++++++++----------
> > > > >  arch/riscv/mm/physaddr.c      | 16 ++++++++++++++++
> > > > >  drivers/of/fdt.c              | 11 ++++++-----
> > > > >  4 files changed, 57 insertions(+), 17 deletions(-)
> > > > >
> > > > > --
> > > > > 2.37.2
> > > > >