[lkml]   [2012]   [Aug]   [28]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: [PATCH v2 1/1] mmc: block: Add write packing control
On Mon, Aug 27, 2012 at 11:58 PM,  <> wrote:
> On Fri, July 27, 2012 2:07 am, S, Venkatraman wrote:
>> On Fri, Jul 27, 2012 at 12:24 AM, <> wrote:
>>> On Thu, July 26, 2012 8:28 am, S, Venkatraman wrote:
>>>> On Tue, Jul 24, 2012 at 2:14 PM, <> wrote:
>>>>> On Mon, July 23, 2012 5:22 am, S, Venkatraman wrote:
>>>>>> On Mon, Jul 23, 2012 at 5:13 PM, <> wrote:
>>>>>>> On Wed, July 18, 2012 12:26 am, Chris Ball wrote:
>>>>>>>> Hi, [removing Jens and the documentation list, since now we're
>>>>> talking about the MMC side only]
>>>>>>>> On Wed, Jul 18 2012, wrote:
>>>>>>>>> Is there anything else that holds this patch from being pushed to
>>>>>>> mmc-next?
>>>>>>>> Yes, I'm still uncomfortable with the write packing patchsets for a
>>>>>>> couple of reasons, and I suspect that the sum of those reasons means
>>>>>>> that
>>>>>>> we should probably plan on holding off merging it until after 3.6.
>>>>>>>> Here are the open issues; please correct any misunderstandings:
>>>>>>>> With
>>>>> Seungwon's patchset ("Support packed write command"):
>>>>>>>> * I still don't have a good set of representative benchmarks
>>>>>>>> showing
>>>>>>>> what kind of performance changes come with this patchset. It
>>>>>>>> seems
>>>>>>> like we've had a small amount of testing on one controller/eMMC part
>>>>>>> combo
>>>>>>> from Seungwon, and an entirely different test from Maya, and the
>>>>> results
>>>>>>> aren't documented fully anywhere to the level of describing what the
>>>>> hardware was, what the test was, and what the results were before and
>>>>> after the patchset.
>>>>>>> Currently, there is only one card vendor that supports packed
>>>>>>> commands.
>>>>> Following are our sequential write (LMDD) test results on 2 of our
>>>>> targets
>>>>>>> (in MB/s):
>>>>>>> No packing packing
>>>>>>> Target 1 (SDR 50MHz) 15 25
>>>>>>> Target 2 (DDR 50MHz) 20 30
>>>>>>>> With the reads-during-writes regression:
>>>>>>>> * Venkat still has open questions about the nature of the read
>>>>>>>> regression, and thinks we should understand it with blktrace
>>>>>>>> before
>>>>>>> trying to fix it. Maya has a theory about writes overwhelming
>>>>>>> reads,
>>>>>>> but
>>>>>>> Venkat doesn't understand why this would explain the observed
>>>>>>> bandwidth drop.
>>>>>>> The degradation of read due to writes is not a new behavior and
>>>>>>> exists
>>>>> also without the write packing feature (which only increases the
>>>>> degradation). Our investigation of this phenomenon led us to the
>>>>> Conclusion that a new scheduling policy should be used for mobile
>>>>> devices,
>>>>>>> but this is not related to the current discussion of the write
>>>>>>> packing
>>>>> feature.
>>>>>>> The write packing feature increases the degradation of read due to
>>>>> write
>>>>>>> since it allows the MMC to fetch many write requests in a row,
>>>>>>> instead
>>>>>>> of
>>>>>>> fetching only one at a time. Therefore some of the read requests
>>>>>>> will
>>>>> have to wait for the completion of more write requests before they can
>>>>> be
>>>>>>> issued.
>>>>>> I am a bit puzzled by this claim. One thing I checked carefully when
>>>>> reviewing write packing patches from SJeon was that the code didn't
>>>>> plough through a mixed list of reads and writes and selected only
>>>>> writes.
>>>>>> This section of the code in "mmc_blk_prep_packed_list()", from v8
>>>>> patchset..
>>>>>> <Quote>
>>>>>> + if (rq_data_dir(cur) != rq_data_dir(next)) {
>>>>>> + put_back = 1;
>>>>>> + break;
>>>>>> + }
>>>>>> </Quote>
>>>>>> means that once a read is encountered in the middle of write packing,
>>>>> the packing is stopped at that point and it is executed. Then the next
>>>>> blk_fetch_request should get the next read and continue as before.
>>>>>> IOW, the ordering of reads and writes is _not_ altered when using
>>>>>> packed
>>>>> commands.
>>>>>> For example if there were 5 write requests, followed by 1 read,
>>>>>> followed by 5 more write requests in the request_queue, the first 5
>>>>> writes will be executed as one "packed command", then the read will be
>>>>> executed, and then the remaining 5 writes will be executed as one
>>>>> "packed command". So the read does not have to wait any more than it
>>>>> waited before (packing feature)
>>>>> Let me try to better explain with your example.
>>>>> Without packing the MMC layer will fetch 2 write requests and wait for
>>>>> the
>>>>> first write request completion before fetching another write request.
>>>>> During this time the read request could be inserted into the CFQ and
>>>>> since
>>>>> it has higher priority than the async write it will be dispatched in
>>>>> the
>>>>> next fetch. So, the result would be 2 write requests followed by one
>>>>> read
>>>>> request and the read would have to wait for completion of only 2 write
>>>>> requests.
>>>>> With packing, all the 5 write requests will be fetched in a row, and
>>>>> then
>>>>> the read will arrive and be dispatched in the next fetch. Then the
>>>>> read
>>>>> will have to wait for the completion of 5 write requests.
>>>>> Few more clarifications:
>>>>> Due to the plug list mechanism in the block layer the applications can
>>>>> "aggregate" several requests to be inserted into the scheduler before
>>>>> waking the MMC queue thread.
>>>>> This leads to a situation where there are several write requests in
>>>>> the
>>>>> CFQ queue when MMC starts to do the fetches.
>>>>> If the read was inserted while we are building the packed command then
>>>>> I
>>>>> agree that we should have seen less effect on the read performance.
>>>>> However, the write packing statistics show that in most of the cases
>>>>> the
>>>>> packing stopped due to an empty queue, meaning that the read was
>>>>> inserted
>>>>> to the CFQ after all the pending write requests were fetched and
>>>>> packed.
>>>>> Following is an example for write packing statistics of a READ/WRITE
>>>>> parallel scenario:
>>>>> write packing statistics:
>>>>> Packed 1 reqs - 448 times
>>>>> Packed 2 reqs - 38 times
>>>>> Packed 3 reqs - 23 times
>>>>> Packed 4 reqs - 30 times
>>>>> Packed 5 reqs - 14 times
>>>>> Packed 6 reqs - 8 times
>>>>> Packed 7 reqs - 4 times
>>>>> Packed 8 reqs - 1 times
>>>>> Packed 10 reqs - 1 times
>>>>> Packed 34 reqs - 1 times
>>>>> stopped packing due to the following reasons:
>>>>> 2 times: wrong data direction (meaning a READ was fetched and stopped
>>>>> the
>>>>> packing)
>>>>> 1 times: flush or discard
>>>>> 565 times: empty queue (meaning blk_fetch_request returned NULL)
>>>>>> And I requested blktrace to confirm that this is indeed the
>>>>>> behaviour.
>>>>> The trace logs show that in case of no packing, there are maximum of
>>>>> 3-4
>>>>> requests issued before a read request, while with packing there are
>>>>> also
>>>>> cases of 6 and 7 requests dispatched before a read request.
>>>>> I'm waiting for an approval for sharing the block trace logs.
>>>>> Since this is a simple test to run you can collect the trace logs and
>>>>> let
>>>>> us know if you reach other conclusions.
>>>> Thanks for the brief. I don't have the eMMC4.5 device with me yet, so
>>>> I can't reproduce the result.
>>> I sent the trace logs of both packing and non packing. Please let me
>>> know
>>> if you have additional questions after reviewing them.
>>> The problem you describe is most likely
>>>> applicable
>>>> to any block device driver with a large queue depth ( any queue depth
>>>> >1).
>>>> I'll check to see what knobs in block affect the result.
>>>> Speaking of it, what is the host controller you use to test this ?
>>> The controller I use is msm_sdcc.
>>>> I was wondering if host->max_seg_size is taken into account while
>>>> packed
>>>> command
>>>> is in use. If not, shouldn't it be ? - it could act as a better
>>>> throttle for "packing density".
>>> The max segments (which is calculated from host->max_seg_size) is taking
>>> into account when preparing the packed list (so that the whole packed
>>> won't exceed the max number of segments).
>>> I'm not sure I understand how host->max_seg_size can be used as a
>>> throttle
>>> for "packing density". Can you please explain?
>> Ok - I overlooked that max_segments is indeed used to limit the number
>> of requests
>> that are packed.(And this corresponds to max_seg_size, which is what I
>> intended)
>> I should be getting my MMC4.5 test gear in a couple of days - I'll run
>> it through
>> on some hosts and can either provide more feedback or Ack this patch.
>> Regards,
>> Venkat.
> Hi Venkat,
> Do you have additional questions/comments?
None. I am just running some stress tests on BKOPS patches right now
and after tomorrow I'll start testing packed command. Will all the
patches apply on top of current mmc-next ? If not, it would be great
you can send an updated version.


 \ /
  Last update: 2012-08-28 20:41    [W:0.083 / U:0.704 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site