Re: Questionable aspects of QEMU Error's design

[Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>]

28.04.2020 08:20, Vladimir Sementsov-Ogievskiy wrote:
> 27.04.2020 18:36, Markus Armbruster wrote:
>> Markus Armbruster <armbru@redhat.com> writes:
>>
>>> Markus Armbruster <armbru@redhat.com> writes:
>>>
>>>> QEMU's Error was patterned after GLib's GError.  Differences include:
>>> [...]
>>>> * Return value conventions
>>>>
>>>>    Common: non-void functions return a distinct error value on failure
>>>>    when such a value can be defined.  Patterns:
>>>>
>>>>    - Functions returning non-null pointers on success return null pointer
>>>>      on failure.
>>>>
>>>>    - Functions returning non-negative integers on success return a
>>>>      negative error code on failure.
>>>>
>>>>    Different: GLib discourages void functions, because these lead to
>>>>    awkward error checking code.  We have tons of them, and tons of
>>>>    awkward error checking code:
>>>>
>>>>      Error *err = NULL;
>>>>      frobnicate(arg, &err);
>>>>      if (err) {
>>>>          ... recover ...
>>>>          error_propagate(errp, err);
>>>>      }
>>>>
>>>>    instead of
>>>>
>>>>      if (!frobnicate(arg, errp))
>>>>          ... recover ...
>>>>      }
>>>>
>>>>    Can also lead to pointless creation of Error objects.
>>>>
>>>>    I consider this a design mistake.  Can we still fix it?  We have more
>>>>    than 2000 void functions taking an Error ** parameter...
>>>>
>>>>    Transforming code that receives and checks for errors with Coccinelle
>>>>    shouldn't be hard.  Transforming code that returns errors seems more
>>>>    difficult.  We need to transform explicit and implicit return to
>>>>    either return true or return false, depending on what we did to the
>>>>    @errp parameter on the way to the return.  Hmm.
>>> [...]
>>>
>>> To figure out what functions with an Error ** parameter return, I used
>>> Coccinelle to find such function definitions and print the return types.
>>> Summary of results:
>>>
>>>     2155 void
>>>      873 signed integer
>>>      494 pointer
>>>      153 bool
>>>       33 unsigned integer
>>>        6 enum
>>>     ---------------------
>>>     3714 total
>>>
>>> I then used Coccinelle to find checked calls of void functions (passing
>>> &error_fatal or &error_abort is not considered "checking" here).  These
>>> calls become simpler if we make the functions return a useful value.  I
>>> found a bit under 600 direct calls, and some 50 indirect calls.
>>>
>>> Most frequent direct calls:
>>>
>>>      127 object_property_set_bool
>>>       27 qemu_opts_absorb_qdict
>>>       16 visit_type_str
>>>       14 visit_type_int
>>>       10 visit_type_uint32
>>>
>>> Let's have a closer look at object_property_set() & friends.  Out of
>>> almost 1000 calls, some 150 are checked.  While I'm sure many of the
>>> unchecked calls can't actually fail, I am concerned some unchecked calls
>>> can.
>>>
>>> If we adopt the convention to return a value that indicates success /
>>> failure, we should consider converting object.h to it sooner rather than
>>> later.
>>>
>>> Please understand these are rough numbers from quick & dirty scripts.
>>
>> FYI, I'm working on converting QemuOpts, QAPI visitors and QOM.  I keep
>> running into bugs.  So far:
>>
>>      [PATCH v2 for-5.1 0/9] qemu-option: Fix corner cases and clean up
>>      [PATCH for-5.1 0/5] qobject: Minor spring cleaning
>>      [PATCH v2 00/14] Miscellaneous error handling fixes
>>      [PATCH 0/4] Subject: [PATCH 0/4] smbus: SPD fixes
>>      [PATCH 0/3] fuzz: Probably there is a better way to do this
>>      [PATCH v2 00/15] qapi: Spring cleaning
>>      [PATCH 00/11] More miscellaneous error handling fixes
>>
>> I got another one coming for QOM and qdev before I can post the
>> conversion.
>>
>> Vladimir, since the conversion will mess with error_propagate(), I'd
>> like to get it in before your auto-propagation work.
>>
> 
> OK, just let me know when to regenerate the series, it's not hard.
> 

Hi! Is all that merged? Should I resend now?

-- 
Best regards,
Vladimir