Re: [linux-next:master 3139/4234] error: <unknown>:0:0: ran out of registers during register allocation in function 'hv_call_get_vp_registers'

[Naman Jain <namjain@linux.microsoft.com>]

On 8/25/2025 10:30 PM, Nathan Chancellor wrote:
> Hi Naman,
> 
> On Sun, Aug 24, 2025 at 11:13:51PM +0530, Naman Jain wrote:
>>
>>
>> On 8/24/2025 12:02 PM, kernel test robot wrote:
>>> tree:   https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master
>>> head:   0f4c93f7eb861acab537dbe94441817a270537bf
>>> commit: 96a1d2495c2f48467a91b8657bd594350b84ea9e [3139/4234] Drivers: hv: Introduce mshv_vtl driver
>>> config: i386-buildonly-randconfig-2003-20250824 (https://download.01.org/0day-ci/archive/20250824/202508240828.pVaUV45D-lkp@intel.com/config)
>>> compiler: clang version 20.1.8 (https://github.com/llvm/llvm-project 87f0227cb60147a26a1eeb4fb06e3b505e9c7261)
>>> reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250824/202508240828.pVaUV45D-lkp@intel.com/reproduce)
>>>
>>> If you fix the issue in a separate patch/commit (i.e. not just a new version of
>>> the same patch/commit), kindly add following tags
>>> | Reported-by: kernel test robot <lkp@intel.com>
>>> | Closes: https://lore.kernel.org/oe-kbuild-all/202508240828.pVaUV45D-lkp@intel.com/
>>>
>>> All errors (new ones prefixed by >>):
>>>
>>>>> error: <unknown>:0:0: ran out of registers during register allocation in function 'hv_call_get_vp_registers'
>>>>> error: <unknown>:0:0: ran out of registers during register allocation in function 'hv_call_set_vp_registers'
>>>>> error: <unknown>:0:0: ran out of registers during register allocation in function 'hv_call_get_partition_property'
>>>      3 errors generated.
>>>
>>
>> +Nathan
>>
>> I can see a bunch of similar warnings previously reported, in different
>> functions, but could not find a fix for it, or if this
>> is a blocking error:
>>
>> https://lore.kernel.org/all/?q=s%3A%22ran+out+of+registers+during+register+allocation%22
>>
>> These all are reported on below 32 bit config, compiled with LLVM:
>> i386-buildonly-randconfig-*
>>
>> Found previous comment from Nathan Chancellor, here, which seems to be
>> related to this error:
>> https://lore.kernel.org/all/20250507022649.GA1984217@ax162/
>>
>> """
>> This configuration has CONFIG_MATOM=y, which does not really work well
>> with clang for this reason and some others that have come up in the
>> past. Given it is likely that nobody will care enough to fix this on the
>> LLVM side (since it is 32-bit), we should probably just block using
>> MATOM with CC_IS_CLANG, which Arnd and I had talked about at the last
>> ClangBuiltLinux meeting.
>> """
>>
>> This gives me an impression that we can safely ignore these errors from
>> kernel test bot.
>> Requesting Nathan to please correct me if I am wrong.
> 
> Yes, this is all correct. I will try to pursue an actual kernel side fix
> for this in Kconfig but I think the Intel folks can also start ignoring
> these reports for ARCH=i386 and CONFIG_MATOM so that they do not get
> sent to other kernel developers for random changes.
> 
> Cheers,
> Nathan

Hi Nathan,
Thanks for confirming.

Regards,
Naman