On 10/21/24 07:36, Guenter Roeck wrote:
On 10/21/24 03:43, BALATON Zoltan wrote:
(Resend with fixed ppc list address.)
On Sun, 20 Oct 2024, Guenter Roeck wrote:
Hi,
On Fri, Jul 19, 2024 at 11:07:01AM +1000, Richard Henderson wrote:
We can determine at translation time whether the insn is or
is not dbczl. We must retain a runtime check against the
HID5 register, but we can move that to a separate function
that never affects other ppc models.
Reviewed-by: Nicholas Piggin <npig...@gmail.com>
Reviewed-by: BALATON Zoltan <bala...@eik.bme.hu>
Signed-off-by: Richard Henderson <richard.hender...@linaro.org>
I see an odd failure when trying to boot ppc64 images in qemu v9.1.0 and
v9.1.1.
Starting network: udhcpc: started, v1.36.1
malloc(): corrupted top size
Aborted
Do you have more info on what reproduces this? What is the QEMU command and
what guest code is running? This looks like something may be using dcbzl on
something else than a 970 or if this is emulating a 970 then I don't see why
this stopped working. Looking at the reproducer may help.
It affects all CPUs I am using for the various 64-bit powerpc tests. Those are
ppc64, POWER8, POWER9, e5500, and power10. I don't see the problem with 32-bit
emulations.
A sample command line is
qemu-system-ppc64 -M mac99 -cpu ppc64 \
-m 1G -kernel vmlinux \
-initrd rootfs.cpio \
-device ne2k_pci,netdev=net0 -netdev user,id=net0 \
-nographic -vga none -monitor null -no-reboot \
--append "rdinit=/sbin/init console=tty console=ttyS0"
with the root file system from
https://github.com/groeck/linux-build-test/blob/master/rootfs/ppc64/rootfs.cpio.gz
Other than that, I'll be happy to set up a complete reproducer with images etc,
but that will take a while. Turns out that qemu v9.1 has a number of problems,
and it will take me some time to dig through all of them.
Never mind, it turned out to be straightforward.
Try https://github.com/groeck/linux-test-downloads/tree/main/ppc64
Thanks,
Guenter
