[SSSD] [PATCH] sss_cache: Remove fastcache even if sssd is not running.

Michal Židek mzidek at redhat.com
Wed Oct 31 11:43:09 UTC 2012


On 10/30/2012 02:25 PM, Simo Sorce wrote:
> On Tue, 2012-10-30 at 10:39 +0100, Michal Židek wrote:
>> On 10/29/2012 03:53 PM, Simo Sorce wrote:
>>>
>>> Codewise looks ok, but I still see duplication of the code used to lock
>>> the file.
>>>
>>> I was wondering, would it make sense to split this patch in 2 and put
>>> the lock_mc_file functions as a more generic function in the common
>>> utils as a separate patch first ?
>>>
>>> I would see it added in util/util_lock.c so it is available to both
>>> tools and responders (avoids the duplication you still have).
>>>
>>> The prototype should be something like:
>>>
>>> static errno_t sss_br_lock_file(int fd, size_t start, size_t len, int
>>> retries, int wait);
>>>
>>> Where wait is expressed in milliseconds.
>>>
>>
>> I was thinking about this too, but there is one thing I do not like
>> about it. Maybe I am wrong, but IMO functions related to memory cache
>> should be as private to sssd_nss as possible. That is why the function
>> for locking the files is static (sssd_nss part). Function for memory
>> cache is invalidation is an exception to this rule, because it is used
>> by sss_cache, but it should also keep the code that manipulates memory
>> cache private (and it uses it only on one place, so static function was
>> not needed that much).
>
> This is why I made it a generic function not strictly related to the
> cache in the prototype.
> I prefer to keep a single function to be used by all components, because
> copy&paste is a source of pain in the long term.
>
>> I do not know about any other place, where we would like to lock the mc
>> files except of sssd_nss and the sss_memcache_invalidate used by
>> sss_cache and it should stay like this. I consider the usage of mc files
>> in sss_cache as necessary pollution, but providing non-static API to
>> spread this pollution does not sound good to me.
>
> The function prototype I suggested is not mc specific. Look at it, it
> can be used with any file to lock any part, we can also extend it to
> take a lock type if you want it even more generic.
>
>> But if you really want the util_lock.c, I can send it as a separate
>> patch on top of this one (it is only a small refactor). But still... I
>> am not sure if this is good idea.
>
> I think it should be a patch that comes before the current one and just
> provides the facility, then the current patch will build on top of it.
>

Ok. I did not add the parameter to select the type of lock (not all 
types would fit to this function). But we can expand the util_lock.c 
later with additional functions if more lock types are needed.

>>> Also you do not really need to unlock the file if you are going to
>>> close() it.
>>> Posix semantics require the OS to drop all locks on a file if you close
>>> it, so you can safely drop the logic around unlocking in tools_util.c
>>> (however add a comment before the close like: /* closing the file will
>>> drop the lock */ ).
>>
>> I thought closing the file explicitly is more readable. But yes, I'll
>> remove it, adding the comment as you suggested is better.
>
> Yup, thanks, this part looks good, I think we only need to split the
> patch and move the locking function into util, then we are golden.
>
> Simo.
>

New patches attached.

Thanks
Michal

-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-util-Added-new-file-util_lock.c.patch
Type: text/x-patch
Size: 4223 bytes
Desc: not available
URL: <https://lists.fedorahosted.org/pipermail/sssd-devel/attachments/20121031/61c56aaa/attachment.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0002-sss_cache-Remove-fastcache-even-if-sssd-is-not-runni.patch
Type: text/x-patch
Size: 9515 bytes
Desc: not available
URL: <https://lists.fedorahosted.org/pipermail/sssd-devel/attachments/20121031/61c56aaa/attachment-0001.bin>


More information about the sssd-devel mailing list