Re: [RFC 1/3] procfs: fdinfo -- Extend information about epoll target files

[Pavel Emelyanov]

On 02/22/2017 10:54 AM, Cyrill Gorcunov wrote:
> On Wed, Feb 22, 2017 at 10:44:07AM +0300, Pavel Emelyanov wrote:
>> On 02/21/2017 10:16 PM, Cyrill Gorcunov wrote:
>>> On Tue, Feb 21, 2017 at 10:41:12AM -0800, Andy Lutomirski wrote:
>>>>> Thus lets add file position, inode and device number where
>>>>> this target lays. This three fields can be used as a primary
>>>>> key for sorting, and together with kcmp help CRIU can find
>>>>> out an exact file target (from the whole set of processes
>>>>> being checkpointed).
>>>>
>>>> I have no problem with this, but I'm wondering whether kcmp's ordered
>>>> comparisons could also be used for this purpose.
>>>
>>> Yes it can, but it would increas number of kcmp calls signisicantly.
>>
>> Actually it shouldn't. If you extend the kcmp argument to accept the
>> epollfd:epollslot pair, this would be effectively the same as if you
>> had all your epoll-ed files injected into your fdtable with "strange"
>> fd numbers. We already have two-level rbtree for this in criu, adding
>> extended ("strange") fd to it should be OK.
> 
> Nope. Pavel, I guess you forget how we handle file tree in criu currently.
> We call for kcmp only if we have to -- when primary key for two entries
> is the same.

True, but the latter is an optimization to reduce the number of syscalls.

Look, in order to have a primary key you need to do some system call for the
fd you check (read from proc or stat the descriptor). But for target files in
e-polls you don't make this per-fd syscall to get primary key, just call the
kcmp instead.

-- Pavel