RFR: JDK-8158502: aarch32: detect QEMU user-mode emulation
Alex Kashchenko
akashche at redhat.com
Mon Jun 20 10:03:42 UTC 2016
Hi,
Commenting on this bit:
On 06/02/2016 05:51 PM, Andrey Petushkov wrote:
> I can assume that the real reason for the problem is absence of
indication of the cpu version in the /proc/cpuinfo, since java sees the
host’s version, which has nothing to do with instruction set being
emulated. As such we fall into lowest armv6 mode, which is actually not
supported by the code
Just wanted to confirm that I've got the same assertions on relocations
running armv6-hardfloat builds (on RPi B+). It still asserts later in
assembler_aarch32 if relocation errors are ignored though (that is
likely expected for armv6).
On 06/08/2016 04:20 PM, Andrey Petushkov wrote:
> Hi Ed!
>
> Cool, thank you, did not know such thing exists! Since it’s available since kernel 3.1 IMHO it’s ok to rely on it. I will give it a try after finish with v6k
>
> Regards,
> Andrey
>
>> On 08 Jun 2016, at 18:05, Edward Nevill <edward.nevill at gmail.com> wrote:
>>
>> Hi,
>>
>> Is it possible to use the kernel user helper functions to implement the
>> required functionality for ldrexd/strexd?
>>
>> https://www.kernel.org/doc/Documentation/arm/kernel_user_helpers.txt
>>
>> I am thinking of __kuser_cmpxchg64
>>
>> Regards,
>> Ed.
>>
>> On Wed, 2016-06-08 at 17:41 +0300, Andrey Petushkov wrote:
>>> Hi Alex,
>>>
>>> Thank you! And you’re seem to be right, ARM1136 seem to be “pure” v6, without extensions brought by v6k. One problem though, it is soft-float as you say. And to my best knowledge our port is not ready for pure soft-float environment. It does not respect soft-float ABI. In addition, I’ve stumbled across one unpleasant problem of missing ldrexd/strext instruction in v6. So I might reconsider support of pure v6, at least for the time being. So please hold off from setting up this machine, until further notice.
>>> At the same time support for v6k remains my first priority so keep your RPi B+ warmed :)
>>> (I've finished the detection code and started battle against relocations)
>>>
>>> Regards,
>>> Andrey
>>>
>>
>>
>
--
-Alex
More information about the aarch32-port-dev
mailing list