Re: [Qemu-devel] [PATCH] qga: ignore non present cpus when handling qmp_guest_get_vcpus()

[Laszlo Ersek <lersek@redhat.com>]

On 09/06/18 12:50, Igor Mammedov wrote:
> On Thu, 6 Sep 2018 12:26:12 +0200
> Laszlo Ersek <lersek@redhat.com> wrote:
> 
>> On 09/06/18 11:49, Igor Mammedov wrote:
>>> On Thu, 30 Aug 2018 17:51:13 +0200
>>> Laszlo Ersek <lersek@redhat.com> wrote:
>>>   
>>>> +Drew
>>>>
>>>> On 08/30/18 14:08, Igor Mammedov wrote:  
>>>>> If VM has VCPUs plugged sparselly (for example a VM started with
>>>>> 3 VCPUs (cpu0, cpu1 and cpu2) and then cpu1 was hotunplugged so
>>>>> only cpu0 and cpu2 are present), QGA will rise a error
>>>>>   error: internal error: unable to execute QEMU agent command 'guest-get-vcpus':
>>>>>   open("/sys/devices/system/cpu/cpu1/"): No such file or directory
>>>>> when
>>>>>   virsh vcpucount FOO --guest
>>>>> is executed.
>>>>> Fix it by ignoring non present CPUs when fetching CPUs status from sysfs.
>>>>>
>>>>> Signed-off-by: Igor Mammedov <imammedo@redhat.com>
>>>>> ---
>>>>>  qga/commands-posix.c | 4 +++-
>>>>>  1 file changed, 3 insertions(+), 1 deletion(-)
>>>>>
>>>>> diff --git a/qga/commands-posix.c b/qga/commands-posix.c
>>>>> index 37e8a2d..2929872 100644
>>>>> --- a/qga/commands-posix.c
>>>>> +++ b/qga/commands-posix.c
>>>>> @@ -2044,7 +2044,9 @@ static void transfer_vcpu(GuestLogicalProcessor *vcpu, bool sys2vcpu,
>>>>>                                vcpu->logical_id);
>>>>>      dirfd = open(dirpath, O_RDONLY | O_DIRECTORY);
>>>>>      if (dirfd == -1) {
>>>>> -        error_setg_errno(errp, errno, "open(\"%s\")", dirpath);
>>>>> +        if (!(sys2vcpu && errno == ENOENT)) {
>>>>> +            error_setg_errno(errp, errno, "open(\"%s\")", dirpath);
>>>>> +        }
>>>>>      } else {
>>>>>          static const char fn[] = "online";
>>>>>          int fd;
>>>>>     
>>> [...]
>>>    
>>>> I wonder if, instead of this patch, we should rework
>>>> qmp_guest_get_vcpus(), to silently skip processors for which this
>>>> dirpath ENOENT condition arises (i.e., return a shorter list of
>>>> GuestLogicalProcessor objects).  
>>> would something like this on top of this patch do?
>>>
>>> diff --git a/qga/commands-posix.c b/qga/commands-posix.c
>>> index 2929872..990bb80 100644
>>> --- a/qga/commands-posix.c
>>> +++ b/qga/commands-posix.c
>>> @@ -2114,12 +2114,14 @@ GuestLogicalProcessorList *qmp_guest_get_vcpus(Error **errp)
>>>          vcpu->logical_id = current++;
>>>          vcpu->has_can_offline = true; /* lolspeak ftw */
>>>          transfer_vcpu(vcpu, true, &local_err);
>>> -
>>> -        entry = g_malloc0(sizeof *entry);
>>> -        entry->value = vcpu;
>>> -
>>> -        *link = entry;
>>> -        link = &entry->next;
>>> +        if (errno == ENOENT) {
>>> +            g_free(vcpu);
>>> +        } else {
>>> +            entry = g_malloc0(sizeof *entry);
>>> +            entry->value = vcpu;
>>> +            *link = entry;
>>> +            link = &entry->next;
>>> +        }
>>>      }
>>>  
>>>      if (local_err == NULL) {
>>>
>>> [...]
>>>   
>>
>> To me that looks like the right approach, but the details should be
>> polished a bit:
>>
>> - After we drop the vcpu object, "local_err" is still set, and would
>> terminate the loop in the next iteration.
> local_error is not set due 'if (!(sys2vcpu && errno == ENOENT))'
> condition in the in the transfer_vcpu().

ah, sorry, you did say this was on top of your original patch, but I had
forgotten the details of that.

> the thing is that in case of vcpu2sys direction ENOENT is hard error.
> 
>> - It seems like ENOENT can indeed only come from openat(), in
>> transfer_vcpu(), however, it would be nice if we could grab the error
>> code from the error object somehow, and not from the "errno" variable. I
>> vaguely recall this is what error classes were originally invented for,
>> but now we just use ERROR_CLASS_GENERIC_ERROR...
> I've checked it and errno is preserved during error_setg_errno() call but
> not saved in Error, so I've dropped that idea.
>  
>> How about this: we could add a boolean output param to transfer_vcpu(),
>> called "fatal". Ignored when the function succeeds. When the function
>> fails (seen from "local_err"), the loop consults "fatal". If the error
>> is fatal, we act as before; otherwise, we drop the vcpu object, release
>> -- and zero out -- "local_err" as well, and continue. I think this is
>> more generic / safer than trying to infer the failure location from the
>> outside.
> It looked more uglier to me, so I've turned to libc style of reporting
> (assuming that g_free() doesn't touch errno ever)
> 
> But if you prefer using extra parameter, I'll respin patch with it.

Michael, what is your preference? I guess I'll be fine both ways.

Thanks,
Laszlo

> 
>> I'm not quite up to date on structured error propagation in QEMU, so
>> take the above with a grain of salt...
>>
>> Thanks,
>> Laszlo
>