[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] bug in new ubfx code
From: |
Richard Henderson |
Subject: |
Re: [Qemu-devel] bug in new ubfx code |
Date: |
Fri, 13 Jan 2017 09:39:34 -0800 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101 Thunderbird/45.6.0 |
On 01/13/2017 09:27 AM, Ard Biesheuvel wrote:
> Hi all,
>
> I tracked down a boot issue I was having with running the kernel under
> AArch64 system emulation to commit
>
> It appears that ubfx is executing incorrectly: the following code
>
> .global _start
> _start:
> mov x1, #0x1124
> ubfx x2, x1, #28, #4
>
> built with
>
> aarch64-linux-gnu-gcc -o /tmp/ubfx /tmp/ubfx.s -nostartfiles
> aarch64-linux-gnu-objcopy -O binary /tmp/ubfx.bin /tmp/ubfx
>
> and executed with
>
> qemu-system-aarch64 -M virt -cpu cortex-a53 -kernel /tmp/ubfx -s -S -nographic
>
> produces the value 0x1124000000000 in x2 (as can be observed via GDB),
> while the expected value is 0
Ho hum. Somehow the relevant patch was corrupted. It has half of
Alex's fix in
http://lists.nongnu.org/archive/html/qemu-devel/2016-12/msg00137.html
but is also missing half. I'll fix it up asap.
r~