Hi Per, minor proofreading, On Tue, Jul 05 2011, Per Forlin wrote: > Documentation about the background and the design of mmc non-blocking. > Host driver guide lines to minimize request preparation over head. guidelines, overhead > > Signed-off-by: Per Forlin <per.forlin@xxxxxxxxxx> > --- > Documentation/mmc/00-INDEX | 2 + > Documentation/mmc/mmc-async-req.txt | 85 +++++++++++++++++++++++++++++++++++ > 2 files changed, 87 insertions(+), 0 deletions(-) > create mode 100644 Documentation/mmc/mmc-async-req.txt > > diff --git a/Documentation/mmc/00-INDEX b/Documentation/mmc/00-INDEX > index 93dd7a7..11bc2cf 100644 > --- a/Documentation/mmc/00-INDEX > +++ b/Documentation/mmc/00-INDEX > @@ -4,3 +4,5 @@ mmc-dev-attrs.txt > - info on SD and MMC device attributes > mmc-dev-parts.txt > - info on SD and MMC device partitions > +mmc-async-req.txt > + - info on mmc asynchronous request > diff --git a/Documentation/mmc/mmc-async-req.txt b/Documentation/mmc/mmc-async-req.txt > new file mode 100644 > index 0000000..d139a51 > --- /dev/null > +++ b/Documentation/mmc/mmc-async-req.txt > @@ -0,0 +1,85 @@ > +Rationale > +========= > + > +How significant is the cache maintenance over head? overhead > +It depends, fast eMMC and multiple cache levels with speculative cache pre-fetch This line is over 80 cols. Please wrap at around 76 cols. > +makes the cache overhead relatively significant. If the DMA preparations > +for the next request is done in parallel to the current transfer are done > +the DMA preparation overhead would not affect the MMC performance. > +The intention of non-blocking (asynchronous) mmc requests is to minimize the > +time between a mmc request ends and another mmc request begins. between when an > +Using mmc_wait_for_req() the MMC controller is idle when dma_map_sg and s/when/while/ > +dma_unmap_sg is processing. Using non-blocking mmc request makes it requests > +possible to prepare the caches for next job in parallel to an active > +mmc request. > + > +MMC block driver > +================ > + > +The issue_rw_rq() in the mmc block driver is made non-blocking. > +The increase in throughput is proportional to the time it takes to > +prepare (major part of preparations is dma_map_sg and dma_unmap_sg) > +a request and how fast the memory is. The faster the MMC/SD is > +the more significant the prepare request time becomes. Roughly the expected > +performance gain is 5% for large writes and 10% on large reads on a L2 cache > +platform. In power save mode, when clocks run on a lower frequency, the DMA > +preparation may cost even more. As long as these slower preparations are run > +in parallel to the transfer performance wont be affected. > + > +Details on measurements from IOZone and mmc_test > +================================================ > + > +https://wiki.linaro.org/WorkingGroups/Kernel/Specs/StoragePerfMMC-async-req > + > +MMC core API extension > +====================== > + > +There is one new public function mmc_start_req() > +Is starts a new MMC command request for a host. The function isn't s/Is/It/ > +truely non-blocking. If there is on ongoing async request it waits truly > +for completion of that request and starts the new one and return. It start > +Doesn't wait for the new request to complete. If there is no ongoing s/Doesn't/doesn't/ > +request it starts the new request and returns immediately. > + > +MMC host extensions > +=================== > + > +There are two optional hooks pre_req() and post_req() that the host driver > +may implement in order to move work to before and after the actual mmc_request > +function is called. In the DMA case pre_req() may do dma_map_sg() and prepare > +the dma descriptor, and post_req runs the dma_unmap_sg. > + > +Optimize for the first request > +============================== > + > +The first request in a series of requests can't be prepared in parallel to the > +previous transfer, since there is no previous request. > +The argument is_first_req in pre_req() indicates that there is no previous > +request. The host driver may optimize for this scenario to minimize > +the performance loss. A way to optimize for this is to split the current > +request in two chunks, prepare the first chunk and start the request, > +and finally prepare the second chunk and start the transfer. > + > +Pseudocode to handle is_first_req scenario with minimal prepare over head: overhead > +if (is_first_req && req->size > threshold) > + /* start MMC transfer for the complete transfer size */ > + mmc_start_command(MMC_CMD_TRANSFER_FULL_SIZE) > + > + /* > + * Begin to prepare DMA while cmd is being processed by MMC. > + * The first chunk of the request should take the same time > + * to prepare as the "MMC process command time". > + * If prepare time exceeds MMC cmd time > + * the transfer is delayed, guesstimate max 4k as first chunk size. > + */ > + prepare_1st_chunk_for_dma(req) > + /* flush pending desc to the DMAC (dmaengine.h) */ > + dma_issue_pending(req->dma_desc); > + > + prepare_2st_chunk_for_dma(req) 2nd, not 2st > + /* > + * The second issue_pending should be called before MMC runs out > + * of the first chunk. If the MMC runs out of the first data chunk before > + * this call, the transfer is delayed. > + */ > + dma_issue_pending(req->dma_desc); Thanks, - Chris. -- Chris Ball <cjb@xxxxxxxxxx> <http://printf.net/> One Laptop Per Child -- To unsubscribe from this list: send the line "unsubscribe linux-mmc" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
