[PATCH v1] RISC-V: take text_mutex during alternative patching

[Heiko Stübner]

Am Mittwoch, 22. Februar 2023, 16:31:25 CET schrieb Andrew Jones:
> On Wed, Feb 22, 2023 at 03:27:43PM +0100, Andrew Jones wrote:
> > On Thu, Feb 16, 2023 at 07:33:55AM -0800, Guenter Roeck wrote:
> > > On 2/15/23 14:00, Heiko Stübner wrote:
> > > [ ... ]
> > > > 
> > > > So now I've also tested Palmer's for-next at
> > > >    commit ec6311919ea6 ("Merge patch series "riscv: Optimize function trace"")
> > > > 
> > > > again with the same variants
> > > > - qemu-riscv32 without zbb
> > > > - qemu-riscv32 with zbb
> > > > - qemu-riscv64 without zbb
> > > > - qemu-riscv64 with zbb
> > > > 
> > > > And all of them booted fine into a nfs-root (debian for riscv64 and a
> > > > buildroot for riscv32).
> > > > 
> > > > I even forced a bug into the zbb code to make sure the patching worked
> > > > correctly (where the kernel failed as expected).
> > > > 
> > > > Qemu-version for me was 7.2.50 (v7.2.0-744-g5a3633929a-dirty)
> > > > 
> > > > I did try the one from Debian-stable (qemu-5.2) but that was too old and
> > > > didn't support Zbb yet.
> > > > 
> > > > One thing of note, the "active" 32bit config I had, somehow didn't produce
> > > > working images and I needed to start a new build using the rv32_defconfig.
> > > > 
> > > > So right now, I'm not sure what more to test though.
> > > > 
> > > 
> > > Another example:
> > > 
> > > - build defconfig
> > > - run
> > >   qemu-system-riscv64 -M virt -m 512M -no-reboot -kernel arch/riscv/boot/Image \
> > >     -snapshot -device virtio-blk-device,drive=d0 -drive file=rootfs.ext2,if=none,id=d0,format=raw \
> > >     -append "root=/dev/vda console=ttyS0,115200 earlycon=uart8250,mmio,0x10000000,115200" \
> > >     -nographic -monitor none
> > > 
> > > With CONFIG_RISCV_ISA_ZBB=y, that results in
> > > 
> > > [    0.818263] /dev/root: Can't open blockdev
> > > [    0.818856] VFS: Cannot open root device "/dev/vda" or unknown-block(0,0): error -6
> > > [    0.819177] Please append a correct "root=" boot option; here are the available partitions:
> > > [    0.819808] fe00           16384 vda
> > > [    0.819944]  driver: virtio_blk
> > > [    0.820534] Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(0,0)
> > > [    0.821101] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 6.2.0-rc8-next-20230216-00002-g80332825e240 #4
> > > [    0.821672] Hardware name: riscv-virtio,qemu (DT)
> > > [    0.822050] Call Trace:
> > > [    0.822427] [<ffffffff800053e4>] dump_backtrace+0x1c/0x24
> > > [    0.822834] [<ffffffff807f90e4>] show_stack+0x2c/0x38
> > > [    0.823085] [<ffffffff80803aea>] dump_stack_lvl+0x3c/0x54
> > > [    0.823351] [<ffffffff80803b16>] dump_stack+0x14/0x1c
> > > [    0.823601] [<ffffffff807f944c>] panic+0x102/0x29e
> > > [    0.823834] [<ffffffff80a015e2>] mount_block_root+0x18c/0x23e
> > > [    0.824148] [<ffffffff80a0187c>] mount_root+0x1e8/0x218
> > > [    0.824398] [<ffffffff80a019ee>] prepare_namespace+0x142/0x184
> > > [    0.824655] [<ffffffff80a01182>] kernel_init_freeable+0x236/0x25a
> > > [    0.824934] [<ffffffff80804602>] kernel_init+0x1e/0x10a
> > > [    0.825201] [<ffffffff80003560>] ret_from_exception+0x0/0x16
> > > [    0.826180] ---[ end Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(0,0) ]---
> > > 
> > > This works fine if CONFIG_RISCV_ISA_ZBB is not enabled.
> > > 
> > > Tested with gcc 11.3, binutils 2.39, qemu v7.2.0 and qemu built from mainline.
> > >
> > 
> > Just to +1 this, I get the same result (unable to mount root fs) with
> > 
> > $QEMU -cpu rv64,zbb=on \
> >         -nographic \
> >         -machine virt \
> >         -kernel $KERNEL \
> >         -append 'root=/dev/vda console=ttyS0' \
> >         -drive file=disk.ext4,format=raw,id=hd0 \
> >         -device virtio-blk-pci,drive=hd0
> > 
> > kernel:   latest riscv-linux/for-next (8658db0a4a0f), defconfig
> > gcc:      riscv-gnu-toolchain (12.1.0)
> > binutils: riscv-gnu-toolchain (2.39)
> > qemu:     latest master (79b677d658d3)
> > 
> > Flipping the QEMU cpu zbb property off allows boot to succeed, i.e. it's
> > not necessary to compile out the CONFIG_RISCV_ISA_ZBB code from the
> > kernel, it's just necessary to avoid using it.
> >
> 
> Looks like something in the strncmp implementation. Only commenting it
> out allows boot to succeed.

and interestingly it seems to be something very specific. I.e. my setup is
nfsroot-based (qemu is "just" another board in my boardfarm) and booting
into an nfs-root works quite nicely.

I guess I need to look into how to get an actual disk-image in there.


> diff --git a/arch/riscv/lib/strncmp.S b/arch/riscv/lib/strncmp.S
> index ee49595075be..0bf03abe1ced 100644
> --- a/arch/riscv/lib/strncmp.S
> +++ b/arch/riscv/lib/strncmp.S
> @@ -9,7 +9,7 @@
>  /* int strncmp(const char *cs, const char *ct, size_t count) */
>  SYM_FUNC_START(strncmp)
> 
> -       ALTERNATIVE("nop", "j strncmp_zbb", 0, RISCV_ISA_EXT_ZBB, CONFIG_RISCV_ISA_ZBB)
> +//     ALTERNATIVE("nop", "j strncmp_zbb", 0, RISCV_ISA_EXT_ZBB, CONFIG_RISCV_ISA_ZBB)
> 
>         /*
>          * Returns
> 
> 
> Thanks,
> drew
>