RISC-V: crash in riscv_get_time()

[Bin Meng]

On Fri, Aug 7, 2020 at 6:53 PM Heinrich Schuchardt <xypron.glpk at gmx.de> wrote:
>
> On 06.08.20 16:40, Bin Meng wrote:
> > On Thu, Aug 6, 2020 at 10:28 PM Heinrich Schuchardt <xypron.glpk at gmx.de> wrote:
> >>
> >> On 06.08.20 13:18, Rick Chen wrote:
> >>> Hi Heinrich
> >>>
> >>>> From: Heinrich Schuchardt [mailto:xypron.glpk at gmx.de]
> >>>> Sent: Thursday, August 06, 2020 6:45 PM
> >>>> To: Bin Meng; Rick Jian-Zhi Chen(陳建志)
> >>>> Cc: U-Boot Mailing List
> >>>> Subject: RISC-V: crash in riscv_get_time()
> >>>>
> >>>> Hello Rick, hello Bin,
> >>>>
> >>>> when I run qemu-riscv64_defconfig using
> >>>>
> >>>> qemu-system-riscv64 -machine virt -m 1G -nographic \
> >>>>         -bios u-boot \
> >>>>         -device virtio-net-device,netdev=net0 \
> >>>>         -netdev user,id=net0,tftp=tftp
> >>>>
> >>>> a crash occurs in riscv_get_time() when executing readq().
> >>>>
> >>>> $ qemu-system-riscv64 --version
> >>>> QEMU emulator version 5.0.0 (Debian 1:5.0-13)
> >>>>
> >>>> Looking at doc/board/emulation/qemu-riscv.rst this should work.
> >>>>
> >>>> Is there something missing in the documentation?
> >>>>
> >>>
> >>> You can try this
> >>> ./qemu-system-riscv64 -nographic -machine virt -bios u-boot -device
> >>> virtio-net-device,netdev=net0 -netdev user,id=net0,tftp=tftp
> >>>
> >>> It can work as below:
> >>>
> >>> U-Boot 2020.10-rc1-00299-g89150a9 (Aug 06 2020 - 19:14:01 +0800)
> >>>
> >>> CPU:   rv64imafdcsu
> >>> Model: riscv-virtio,qemu
> >>> DRAM:  128 MiB
> >>> In:    uart at 10000000
> >>> Out:   uart at 10000000
> >>> Err:   uart at 10000000
> >>> Net:   eth0: virtio-net#0
> >>> Hit any key to stop autoboot:  0
> >>> => help
> >>
> >> But on Debian Bullseye it fails:
> >>
> >> $ git reset --hard origin/master
> >> $ make mrproper
> >> $ make qemu-riscv64_defconfig
> >> $ export CROSS_COMPILE
> >> $ export CROSS_COMPILE=riscv64-linux-gnu-
> >> $ qemu-system-riscv64 -nographic -machine virt \
> >>> -bios u-boot \
> >>> -device virtio-net-device,netdev=net0 \
> >>> -netdev user,id=net0,tftp=tftp
> >>
> >
> > Just rebuilt a QEMU 5.0.0 on Ubuntu 16.04 and cannot reproduce.
> > Something is wrong with Debian?
> >
> > Regards,
> > Bin
> >
> I can confirm that with an unpatched QEMU v5.0.0 no problem occurs.
> Same with QEMU 5.0.93 (v5.1.0-rc3-dirty).

Unpatched QEMU? Do you mean Debian shipped a patched QEMU? If so, what
is patched?

Regards,
Bin