From 54d6b44a5f0da708e8a9e496df304e05e8748ef9 Mon Sep 17 00:00:00 2001 From: Pawel Moll Date: Sun, 6 Feb 2011 15:06:24 -0500 Subject: mmc: mmc_test: Only warn about not waiting for busy if it's supported If the MMC host controller does not support waiting for card signaling busy state (MMC_CAP_WAIT_WHILE_BUSY cap), there is no point in prining the relevant warning message. Signed-off-by: Pawel Moll Reviewed-by: Matt Fleming Signed-off-by: Chris Ball --- drivers/mmc/card/mmc_test.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) (limited to 'drivers/mmc/card/mmc_test.c') diff --git a/drivers/mmc/card/mmc_test.c b/drivers/mmc/card/mmc_test.c index 21adc27f4132..0131c6c0d8f6 100644 --- a/drivers/mmc/card/mmc_test.c +++ b/drivers/mmc/card/mmc_test.c @@ -226,9 +226,10 @@ static int mmc_test_wait_busy(struct mmc_test_card *test) if (!busy && mmc_test_busy(&cmd)) { busy = 1; - printk(KERN_INFO "%s: Warning: Host did not " - "wait for busy state to end.\n", - mmc_hostname(test->card->host)); + if (test->card->host->caps & MMC_CAP_WAIT_WHILE_BUSY) + printk(KERN_INFO "%s: Warning: Host did not " + "wait for busy state to end.\n", + mmc_hostname(test->card->host)); } } while (mmc_test_busy(&cmd)); -- cgit From 0532ff6358ae00615cfba7212f5075356b437c66 Mon Sep 17 00:00:00 2001 From: Adrian Hunter Date: Tue, 8 Feb 2011 13:41:01 +0200 Subject: mmc: mmc_test: make performance test area size about 4MiB The test area size was set to the preferred erase size but for comparison purposes it is better if it is the same size for different devices. Make it a multiple of preferred erase size that is greater than or equal to 4MiB. Signed-off-by: Adrian Hunter Signed-off-by: Chris Ball --- drivers/mmc/card/mmc_test.c | 25 +++++++++++++------------ 1 file changed, 13 insertions(+), 12 deletions(-) (limited to 'drivers/mmc/card/mmc_test.c') diff --git a/drivers/mmc/card/mmc_test.c b/drivers/mmc/card/mmc_test.c index 0131c6c0d8f6..ec6060c06ce8 100644 --- a/drivers/mmc/card/mmc_test.c +++ b/drivers/mmc/card/mmc_test.c @@ -1426,28 +1426,29 @@ static int mmc_test_area_cleanup(struct mmc_test_card *test) } /* - * Initialize an area for testing large transfers. The size of the area is the - * preferred erase size which is a good size for optimal transfer speed. Note - * that is typically 4MiB for modern cards. The test area is set to the middle - * of the card because cards may have different charateristics at the front - * (for FAT file system optimization). Optionally, the area is erased (if the - * card supports it) which may improve write performance. Optionally, the area - * is filled with data for subsequent read tests. + * Initialize an area for testing large transfers. The test area is set to the + * middle of the card because cards may have different charateristics at the + * front (for FAT file system optimization). Optionally, the area is erased + * (if the card supports it) which may improve write performance. Optionally, + * the area is filled with data for subsequent read tests. */ static int mmc_test_area_init(struct mmc_test_card *test, int erase, int fill) { struct mmc_test_area *t = &test->area; - unsigned long min_sz = 64 * 1024; + unsigned long min_sz = 64 * 1024, sz; int ret; ret = mmc_test_set_blksize(test, 512); if (ret) return ret; - if (test->card->pref_erase > TEST_AREA_MAX_SIZE >> 9) - t->max_sz = TEST_AREA_MAX_SIZE; - else - t->max_sz = (unsigned long)test->card->pref_erase << 9; + /* Make the test area size about 4MiB */ + sz = (unsigned long)test->card->pref_erase << 9; + t->max_sz = sz; + while (t->max_sz < 4 * 1024 * 1024) + t->max_sz += sz; + while (t->max_sz > TEST_AREA_MAX_SIZE && t->max_sz > sz) + t->max_sz -= sz; t->max_segs = test->card->host->max_segs; t->max_seg_sz = test->card->host->max_seg_size; -- cgit From b6056d12342f936256cbf64c86459af06f11cd5e Mon Sep 17 00:00:00 2001 From: Adrian Hunter Date: Tue, 8 Feb 2011 13:41:02 +0200 Subject: mmc: mmc_test: add tests to measure random I/O operations per second Existing performance tests measure single or sequential I/O speed. Add two random I/O tests: 33. Random read performance by transfer size 34. Random write performance by transfer size Signed-off-by: Adrian Hunter Signed-off-by: Chris Ball --- drivers/mmc/card/mmc_test.c | 139 ++++++++++++++++++++++++++++++++++++++++---- 1 file changed, 128 insertions(+), 11 deletions(-) (limited to 'drivers/mmc/card/mmc_test.c') diff --git a/drivers/mmc/card/mmc_test.c b/drivers/mmc/card/mmc_test.c index ec6060c06ce8..d1aa57a4d058 100644 --- a/drivers/mmc/card/mmc_test.c +++ b/drivers/mmc/card/mmc_test.c @@ -88,6 +88,7 @@ struct mmc_test_area { * @sectors: amount of sectors to check in one group * @ts: time values of transfer * @rate: calculated transfer rate + * @iops: I/O operations per second (times 100) */ struct mmc_test_transfer_result { struct list_head link; @@ -95,6 +96,7 @@ struct mmc_test_transfer_result { unsigned int sectors; struct timespec ts; unsigned int rate; + unsigned int iops; }; /** @@ -495,7 +497,7 @@ static unsigned int mmc_test_rate(uint64_t bytes, struct timespec *ts) */ static void mmc_test_save_transfer_result(struct mmc_test_card *test, unsigned int count, unsigned int sectors, struct timespec ts, - unsigned int rate) + unsigned int rate, unsigned int iops) { struct mmc_test_transfer_result *tr; @@ -510,6 +512,7 @@ static void mmc_test_save_transfer_result(struct mmc_test_card *test, tr->sectors = sectors; tr->ts = ts; tr->rate = rate; + tr->iops = iops; list_add_tail(&tr->link, &test->gr->tr_lst); } @@ -520,20 +523,22 @@ static void mmc_test_save_transfer_result(struct mmc_test_card *test, static void mmc_test_print_rate(struct mmc_test_card *test, uint64_t bytes, struct timespec *ts1, struct timespec *ts2) { - unsigned int rate, sectors = bytes >> 9; + unsigned int rate, iops, sectors = bytes >> 9; struct timespec ts; ts = timespec_sub(*ts2, *ts1); rate = mmc_test_rate(bytes, &ts); + iops = mmc_test_rate(100, &ts); /* I/O ops per sec x 100 */ printk(KERN_INFO "%s: Transfer of %u sectors (%u%s KiB) took %lu.%09lu " - "seconds (%u kB/s, %u KiB/s)\n", + "seconds (%u kB/s, %u KiB/s, %u.%02u IOPS)\n", mmc_hostname(test->card->host), sectors, sectors >> 1, (sectors & 1 ? ".5" : ""), (unsigned long)ts.tv_sec, - (unsigned long)ts.tv_nsec, rate / 1000, rate / 1024); + (unsigned long)ts.tv_nsec, rate / 1000, rate / 1024, + iops / 100, iops % 100); - mmc_test_save_transfer_result(test, 1, sectors, ts, rate); + mmc_test_save_transfer_result(test, 1, sectors, ts, rate, iops); } /* @@ -543,22 +548,24 @@ static void mmc_test_print_avg_rate(struct mmc_test_card *test, uint64_t bytes, unsigned int count, struct timespec *ts1, struct timespec *ts2) { - unsigned int rate, sectors = bytes >> 9; + unsigned int rate, iops, sectors = bytes >> 9; uint64_t tot = bytes * count; struct timespec ts; ts = timespec_sub(*ts2, *ts1); rate = mmc_test_rate(tot, &ts); + iops = mmc_test_rate(count * 100, &ts); /* I/O ops per sec x 100 */ printk(KERN_INFO "%s: Transfer of %u x %u sectors (%u x %u%s KiB) took " - "%lu.%09lu seconds (%u kB/s, %u KiB/s)\n", + "%lu.%09lu seconds (%u kB/s, %u KiB/s, " + "%u.%02u IOPS)\n", mmc_hostname(test->card->host), count, sectors, count, sectors >> 1, (sectors & 1 ? ".5" : ""), (unsigned long)ts.tv_sec, (unsigned long)ts.tv_nsec, - rate / 1000, rate / 1024); + rate / 1000, rate / 1024, iops / 100, iops % 100); - mmc_test_save_transfer_result(test, count, sectors, ts, rate); + mmc_test_save_transfer_result(test, count, sectors, ts, rate, iops); } /* @@ -1768,6 +1775,102 @@ static int mmc_test_profile_seq_trim_perf(struct mmc_test_card *test) return 0; } +static unsigned int rnd_next = 1; + +static unsigned int mmc_test_rnd_num(unsigned int rnd_cnt) +{ + uint64_t r; + + rnd_next = rnd_next * 1103515245 + 12345; + r = (rnd_next >> 16) & 0x7fff; + return (r * rnd_cnt) >> 15; +} + +static int mmc_test_rnd_perf(struct mmc_test_card *test, int write, int print, + unsigned long sz) +{ + unsigned int dev_addr, cnt, rnd_addr, range1, range2, last_ea = 0, ea; + unsigned int ssz; + struct timespec ts1, ts2, ts; + int ret; + + ssz = sz >> 9; + + rnd_addr = mmc_test_capacity(test->card) / 4; + range1 = rnd_addr / test->card->pref_erase; + range2 = range1 / ssz; + + getnstimeofday(&ts1); + for (cnt = 0; cnt < UINT_MAX; cnt++) { + getnstimeofday(&ts2); + ts = timespec_sub(ts2, ts1); + if (ts.tv_sec >= 10) + break; + ea = mmc_test_rnd_num(range1); + if (ea == last_ea) + ea -= 1; + last_ea = ea; + dev_addr = rnd_addr + test->card->pref_erase * ea + + ssz * mmc_test_rnd_num(range2); + ret = mmc_test_area_io(test, sz, dev_addr, write, 0, 0); + if (ret) + return ret; + } + if (print) + mmc_test_print_avg_rate(test, sz, cnt, &ts1, &ts2); + return 0; +} + +static int mmc_test_random_perf(struct mmc_test_card *test, int write) +{ + unsigned int next; + unsigned long sz; + int ret; + + for (sz = 512; sz < test->area.max_tfr; sz <<= 1) { + /* + * When writing, try to get more consistent results by running + * the test twice with exactly the same I/O but outputting the + * results only for the 2nd run. + */ + if (write) { + next = rnd_next; + ret = mmc_test_rnd_perf(test, write, 0, sz); + if (ret) + return ret; + rnd_next = next; + } + ret = mmc_test_rnd_perf(test, write, 1, sz); + if (ret) + return ret; + } + sz = test->area.max_tfr; + if (write) { + next = rnd_next; + ret = mmc_test_rnd_perf(test, write, 0, sz); + if (ret) + return ret; + rnd_next = next; + } + return mmc_test_rnd_perf(test, write, 1, sz); +} + +/* + * Random read performance by transfer size. + */ +static int mmc_test_random_read_perf(struct mmc_test_card *test) +{ + return mmc_test_random_perf(test, 0); +} + +/* + * Random write performance by transfer size. + */ +static int mmc_test_random_write_perf(struct mmc_test_card *test) +{ + return mmc_test_random_perf(test, 1); +} + static const struct mmc_test_case mmc_test_cases[] = { { .name = "Basic write (no data verification)", @@ -2007,6 +2110,20 @@ static const struct mmc_test_case mmc_test_cases[] = { .cleanup = mmc_test_area_cleanup, }, + { + .name = "Random read performance by transfer size", + .prepare = mmc_test_area_prepare, + .run = mmc_test_random_read_perf, + .cleanup = mmc_test_area_cleanup, + }, + + { + .name = "Random write performance by transfer size", + .prepare = mmc_test_area_prepare, + .run = mmc_test_random_write_perf, + .cleanup = mmc_test_area_cleanup, + }, + }; static DEFINE_MUTEX(mmc_test_lock); @@ -2150,11 +2267,11 @@ static int mtf_test_show(struct seq_file *sf, void *data) seq_printf(sf, "Test %d: %d\n", gr->testcase + 1, gr->result); list_for_each_entry(tr, &gr->tr_lst, link) { - seq_printf(sf, "%u %d %lu.%09lu %u\n", + seq_printf(sf, "%u %d %lu.%09lu %u %u.%02u\n", tr->count, tr->sectors, (unsigned long)tr->ts.tv_sec, (unsigned long)tr->ts.tv_nsec, - tr->rate); + tr->rate, tr->iops / 100, tr->iops % 100); } } -- cgit From a803d551d92079beb599cc37fa407c059c82c821 Mon Sep 17 00:00:00 2001 From: Adrian Hunter Date: Tue, 8 Feb 2011 13:41:03 +0200 Subject: mmc: mmc_test: add tests to measure large sequential I/O performance Add two large sequential I/O performance tests: 35. Large sequential read into scattered pages 36. Large sequential write from scattered pages The tests measure transfer times for 10MiB, 100MiB, 1000MiB. Signed-off-by: Adrian Hunter Signed-off-by: Chris Ball --- drivers/mmc/card/mmc_test.c | 100 ++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 100 insertions(+) (limited to 'drivers/mmc/card/mmc_test.c') diff --git a/drivers/mmc/card/mmc_test.c b/drivers/mmc/card/mmc_test.c index d1aa57a4d058..5ec8eddfcf6e 100644 --- a/drivers/mmc/card/mmc_test.c +++ b/drivers/mmc/card/mmc_test.c @@ -1871,6 +1871,92 @@ static int mmc_test_random_write_perf(struct mmc_test_card *test) return mmc_test_random_perf(test, 1); } +static int mmc_test_seq_perf(struct mmc_test_card *test, int write, + unsigned int tot_sz, int max_scatter) +{ + unsigned int dev_addr, i, cnt, sz, ssz; + struct timespec ts1, ts2, ts; + int ret; + + sz = test->area.max_tfr; + /* + * In the case of a maximally scattered transfer, the maximum transfer + * size is further limited by using PAGE_SIZE segments. + */ + if (max_scatter) { + struct mmc_test_area *t = &test->area; + unsigned long max_tfr; + + if (t->max_seg_sz >= PAGE_SIZE) + max_tfr = t->max_segs * PAGE_SIZE; + else + max_tfr = t->max_segs * t->max_seg_sz; + if (sz > max_tfr) + sz = max_tfr; + } + + ssz = sz >> 9; + dev_addr = mmc_test_capacity(test->card) / 4; + if (tot_sz > dev_addr << 9) + tot_sz = dev_addr << 9; + cnt = tot_sz / sz; + dev_addr &= 0xffff0000; /* Round to 64MiB boundary */ + + getnstimeofday(&ts1); + for (i = 0; i < cnt; i++) { + ret = mmc_test_area_io(test, sz, dev_addr, write, + max_scatter, 0); + if (ret) + return ret; + dev_addr += ssz; + } + getnstimeofday(&ts2); + + ts = timespec_sub(ts2, ts1); + mmc_test_print_avg_rate(test, sz, cnt, &ts1, &ts2); + + return 0; +} + +static int mmc_test_large_seq_perf(struct mmc_test_card *test, int write) +{ + int ret, i; + + for (i = 0; i < 10; i++) { + ret = mmc_test_seq_perf(test, write, 10 * 1024 * 1024, 1); + if (ret) + return ret; + } + for (i = 0; i < 5; i++) { + ret = mmc_test_seq_perf(test, write, 100 * 1024 * 1024, 1); + if (ret) + return ret; + } + for (i = 0; i < 3; i++) { + ret = mmc_test_seq_perf(test, write, 1000 * 1024 * 1024, 1); + if (ret) + return ret; + } + + return ret; +} + +/* + * Large sequential read performance. + */ +static int mmc_test_large_seq_read_perf(struct mmc_test_card *test) +{ + return mmc_test_large_seq_perf(test, 0); +} + +/* + * Large sequential write performance. + */ +static int mmc_test_large_seq_write_perf(struct mmc_test_card *test) +{ + return mmc_test_large_seq_perf(test, 1); +} + static const struct mmc_test_case mmc_test_cases[] = { { .name = "Basic write (no data verification)", @@ -2124,6 +2210,20 @@ static const struct mmc_test_case mmc_test_cases[] = { .cleanup = mmc_test_area_cleanup, }, + { + .name = "Large sequential read into scattered pages", + .prepare = mmc_test_area_prepare, + .run = mmc_test_large_seq_read_perf, + .cleanup = mmc_test_area_cleanup, + }, + + { + .name = "Large sequential write from scattered pages", + .prepare = mmc_test_area_prepare, + .run = mmc_test_large_seq_write_perf, + .cleanup = mmc_test_area_cleanup, + }, + }; static DEFINE_MUTEX(mmc_test_lock); -- cgit From 5a8fba524c2063cb297bdbacda535448ee365dc4 Mon Sep 17 00:00:00 2001 From: Chris Ball Date: Wed, 16 Mar 2011 17:46:45 -0400 Subject: mmc: mmc_test: Remove set-but-unused variable. MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Fixes: drivers/mmc/card/mmc_test.c: In function ‘mmc_test_seq_perf’: drivers/mmc/card/mmc_test.c:1878:28: warning: variable ‘ts’ set but not used [-Wunused-but-set-variable] There's no reason to be calling timespec_sub() here, because mmc_test_print_avg_rate() is going to do that itself. Signed-off-by: Chris Ball Cc: Adrian Hunter --- drivers/mmc/card/mmc_test.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'drivers/mmc/card/mmc_test.c') diff --git a/drivers/mmc/card/mmc_test.c b/drivers/mmc/card/mmc_test.c index 5ec8eddfcf6e..f5cedeccad42 100644 --- a/drivers/mmc/card/mmc_test.c +++ b/drivers/mmc/card/mmc_test.c @@ -1875,7 +1875,7 @@ static int mmc_test_seq_perf(struct mmc_test_card *test, int write, unsigned int tot_sz, int max_scatter) { unsigned int dev_addr, i, cnt, sz, ssz; - struct timespec ts1, ts2, ts; + struct timespec ts1, ts2; int ret; sz = test->area.max_tfr; @@ -1912,7 +1912,6 @@ static int mmc_test_seq_perf(struct mmc_test_card *test, int write, } getnstimeofday(&ts2); - ts = timespec_sub(ts2, ts1); mmc_test_print_avg_rate(test, sz, cnt, &ts1, &ts2); return 0; -- cgit