Re: Commit 597ce1723 (MIPS: Support for 64-bit FP with O32 binaries) causing qemu hang with mips64 kernels

[Guenter Roeck <linux@roeck-us.net>]

On 02/07/2014 08:55 PM, Guenter Roeck wrote:
> On 02/07/2014 07:57 PM, Huacai Chen wrote:
>> Hi,
>>
>> Maybe you can try this:
>> http://patchwork.linux-mips.org/patch/6506/
>> http://patchwork.linux-mips.org/patch/6507/
>>
> With those two patches applied the problem is gone.
>

Any idea if and when those patches may find their way upstream ?
3.14-rc4 still has the problem.

Thanks,
Guenter

> Thanks a lot!
>
> Guenter
>
>> On Sat, Feb 8, 2014 at 11:39 AM, Guenter Roeck <linux@roeck-us.net> wrote:
>>>
>>> Hi all,
>>>
>>> in 3.14-rc1, my mips64 qemu test fails. The image boots and then hangs.
>>> Bisect points to commit 597ce1723 (MIPS: Support for 64-bit FP with O32
>>> binaries).
>>> Reverting this commit fixes the problem. Disabling MIPS_O32_FP64_SUPPORT
>>> in my test image does _not_ solve the problem. The qemu version does not
>>> seem
>>> to make a difference; I tested with qemu 1.6.0 and 1.7.0.
>>>
>>> Console log output is available in
>>> http://server.roeck-us.net:8010/builders/qemu-mips64-master/builds/34/steps/buildcommand/logs/stdio
>>>
>>> When the problem is seen, the emulation hangs as can be seen in the log,
>>> and the qemu process consumes 100% CPU until it is killed.
>>>
>>> qemu command line is
>>>
>>> qemu-system-mips64 -kernel vmlinux -M malta -hda
>>> core-image-minimal-qemumips64.ext3 \
>>> -vga cirrus -usb -usbdevice wacom-tablet -no-reboot -m 128 --append
>>> "root=/dev/hda \
>>> rw mem=128M console=ttyS0 console=tty" -nographic
>>>
>>> The same configuration works fine with earlier kernels. I'll be happy to
>>> provide
>>> the detailed configuration as well as the root file system for testing if
>>> needed.
>>>
>>> Obviously I have no idea if this is a problem with the patch or with qemu.
>>> If there is anything I can do to help tracking down the problem further,
>>> please let me know.
>>>
>>> Thanks,
>>> Guenter
>>>
>>
>>
>