On 03/17/2019 02:27 PM, Manfred Spraul wrote:
> Hi Waiman,
>
> On 2/28/19 7:47 PM, Waiman Long wrote:
>> For ipcmni_extend mode, the sequence number space is only 7 bits. So
>> the chance of id reuse is relatively high compared with the non-extended
>> mode.
>>
>> To alleviate this id reuse problem, the id allocation will be done
>> cyclically to cycle through all the 24-bit id space before wrapping
>> around when in ipcmni_extend mode. This may cause the use of more memory
>> in term of the number of xa_nodes allocated as well as potentially more
>> cachelines used as the xa_nodes may be spread more sparsely in this
>> case.
>>
>> There is probably a slight memory and performance cost in doing cyclic
>> id allocation. For applications that really need more than 32k unique
>> IPC
>> identifiers, this is a small price to pay to avoid the id reuse problem.
>
> Have you measured it?
>
> I have observed -3% for semop() for a 4 level radix tree compared to a
> 1-level radix tree, and I'm a bit reluctant to accept that.
> Especially as the percentage will increase if the syscall overhead
> goes down again (-> less spectre impact).
>
It is both Spectre (retpoline) and Meltdown (PTI). PTI is not needed in
AMD CPU and so you may see a bit higher slowdown.
> [...]
>
>> --- a/ipc/util.c
>> +++ b/ipc/util.c
>> @@ -221,7 +221,12 @@ static inline int ipc_idr_alloc(struct ipc_ids
>> *ids, struct kern_ipc_perm *new)
>> */
>> if (next_id < 0) { /* !CHECKPOINT_RESTORE or next_id is unset */
>> - idx = idr_alloc(&ids->ipcs_idr, new, 0, 0, GFP_NOWAIT);
>> + if (ipc_mni_extended)
>> + idx = idr_alloc_cyclic(&ids->ipcs_idr, new, 0, ipc_mni,
>> + GFP_NOWAIT);
>> + else
>> + idx = idr_alloc(&ids->ipcs_idr, new, 0, 0, GFP_NOWAIT);
>> +
>> if ((idx <= ids->last_idx) && (++ids->seq > IPCID_SEQ_MAX))
>> ids->seq = 0;
>
> I don't like it that there are two different codepaths.
>
> Attached is a different proposal:
>
> Always use cyclic allocation, with some logic to minimize the
> additional radix tree levels.
>
> What do you think?
Your proposed patch look good. I saw that you use the max_idx to limit
radix tree nesting level which mitigate my concern of using more memory
and slower performance. I do have some minor comments about the patch in
a later email.
-Longman
