zram: user per-cpu compression streams

Remove idle streams list and keep compression streams in per-cpu data.
This removes two contented spin_lock()/spin_unlock() calls from write
path and also prevent write OP from being preempted while holding the
compression stream, which can cause slow downs.

For instance, let's assume that we have N cpus and N-2
max_comp_streams.TASK1 owns the last idle stream, TASK2-TASK3 come in
with the write requests:

  TASK1            TASK2              TASK3
 zram_bvec_write()
  spin_lock
  find stream
  spin_unlock

  compress

  <<preempted>>   zram_bvec_write()
                   spin_lock
                   find stream
                   spin_unlock
                     no_stream
                       schedule
                                     zram_bvec_write()
                                      spin_lock
                                      find_stream
                                      spin_unlock
                                        no_stream
                                          schedule
   spin_lock
   release stream
   spin_unlock
     wake up TASK2

not only TASK2 and TASK3 will not get the stream, TASK1 will be
preempted in the middle of its operation; while we would prefer it to
finish compression and release the stream.

Test environment: x86_64, 4 CPU box, 3G zram, lzo

The following fio tests were executed:
      read, randread, write, randwrite, rw, randrw
with the increasing number of jobs from 1 to 10.

                  4 streams        8 streams       per-cpu
  ===========================================================
  jobs1
  READ:           2520.1MB/s       2566.5MB/s      2491.5MB/s
  READ:           2102.7MB/s       2104.2MB/s      2091.3MB/s
  WRITE:          1355.1MB/s       1320.2MB/s      1378.9MB/s
  WRITE:          1103.5MB/s       1097.2MB/s      1122.5MB/s
  READ:           434013KB/s       435153KB/s      439961KB/s
  WRITE:          433969KB/s       435109KB/s      439917KB/s
  READ:           403166KB/s       405139KB/s      403373KB/s
  WRITE:          403223KB/s       405197KB/s      403430KB/s
  jobs2
  READ:           7958.6MB/s       8105.6MB/s      8073.7MB/s
  READ:           6864.9MB/s       6989.8MB/s      7021.8MB/s
  WRITE:          2438.1MB/s       2346.9MB/s      3400.2MB/s
  WRITE:          1994.2MB/s       1990.3MB/s      2941.2MB/s
  READ:           981504KB/s       973906KB/s      1018.8MB/s
  WRITE:          981659KB/s       974060KB/s      1018.1MB/s
  READ:           937021KB/s       938976KB/s      987250KB/s
  WRITE:          934878KB/s       936830KB/s      984993KB/s
  jobs3
  READ:           13280MB/s        13553MB/s       13553MB/s
  READ:           11534MB/s        11785MB/s       11755MB/s
  WRITE:          3456.9MB/s       3469.9MB/s      4810.3MB/s
  WRITE:          3029.6MB/s       3031.6MB/s      4264.8MB/s
  READ:           1363.8MB/s       1362.6MB/s      1448.9MB/s
  WRITE:          1361.9MB/s       1360.7MB/s      1446.9MB/s
  READ:           1309.4MB/s       1310.6MB/s      1397.5MB/s
  WRITE:          1307.4MB/s       1308.5MB/s      1395.3MB/s
  jobs4
  READ:           20244MB/s        20177MB/s       20344MB/s
  READ:           17886MB/s        17913MB/s       17835MB/s
  WRITE:          4071.6MB/s       4046.1MB/s      6370.2MB/s
  WRITE:          3608.9MB/s       3576.3MB/s      5785.4MB/s
  READ:           1824.3MB/s       1821.6MB/s      1997.5MB/s
  WRITE:          1819.8MB/s       1817.4MB/s      1992.5MB/s
  READ:           1765.7MB/s       1768.3MB/s      1937.3MB/s
  WRITE:          1767.5MB/s       1769.1MB/s      1939.2MB/s
  jobs5
  READ:           18663MB/s        18986MB/s       18823MB/s
  READ:           16659MB/s        16605MB/s       16954MB/s
  WRITE:          3912.4MB/s       3888.7MB/s      6126.9MB/s
  WRITE:          3506.4MB/s       3442.5MB/s      5519.3MB/s
  READ:           1798.2MB/s       1746.5MB/s      1935.8MB/s
  WRITE:          1792.7MB/s       1740.7MB/s      1929.1MB/s
  READ:           1727.6MB/s       1658.2MB/s      1917.3MB/s
  WRITE:          1726.5MB/s       1657.2MB/s      1916.6MB/s
  jobs6
  READ:           21017MB/s        20922MB/s       21162MB/s
  READ:           19022MB/s        19140MB/s       18770MB/s
  WRITE:          3968.2MB/s       4037.7MB/s      6620.8MB/s
  WRITE:          3643.5MB/s       3590.2MB/s      6027.5MB/s
  READ:           1871.8MB/s       1880.5MB/s      2049.9MB/s
  WRITE:          1867.8MB/s       1877.2MB/s      2046.2MB/s
  READ:           1755.8MB/s       1710.3MB/s      1964.7MB/s
  WRITE:          1750.5MB/s       1705.9MB/s      1958.8MB/s
  jobs7
  READ:           21103MB/s        20677MB/s       21482MB/s
  READ:           18522MB/s        18379MB/s       19443MB/s
  WRITE:          4022.5MB/s       4067.4MB/s      6755.9MB/s
  WRITE:          3691.7MB/s       3695.5MB/s      5925.6MB/s
  READ:           1841.5MB/s       1933.9MB/s      2090.5MB/s
  WRITE:          1842.7MB/s       1935.3MB/s      2091.9MB/s
  READ:           1832.4MB/s       1856.4MB/s      1971.5MB/s
  WRITE:          1822.3MB/s       1846.2MB/s      1960.6MB/s
  jobs8
  READ:           20463MB/s        20194MB/s       20862MB/s
  READ:           18178MB/s        17978MB/s       18299MB/s
  WRITE:          4085.9MB/s       4060.2MB/s      7023.8MB/s
  WRITE:          3776.3MB/s       3737.9MB/s      6278.2MB/s
  READ:           1957.6MB/s       1944.4MB/s      2109.5MB/s
  WRITE:          1959.2MB/s       1946.2MB/s      2111.4MB/s
  READ:           1900.6MB/s       1885.7MB/s      2082.1MB/s
  WRITE:          1896.2MB/s       1881.4MB/s      2078.3MB/s
  jobs9
  READ:           19692MB/s        19734MB/s       19334MB/s
  READ:           17678MB/s        18249MB/s       17666MB/s
  WRITE:          4004.7MB/s       4064.8MB/s      6990.7MB/s
  WRITE:          3724.7MB/s       3772.1MB/s      6193.6MB/s
  READ:           1953.7MB/s       1967.3MB/s      2105.6MB/s
  WRITE:          1953.4MB/s       1966.7MB/s      2104.1MB/s
  READ:           1860.4MB/s       1897.4MB/s      2068.5MB/s
  WRITE:          1858.9MB/s       1895.9MB/s      2066.8MB/s
  jobs10
  READ:           19730MB/s        19579MB/s       19492MB/s
  READ:           18028MB/s        18018MB/s       18221MB/s
  WRITE:          4027.3MB/s       4090.6MB/s      7020.1MB/s
  WRITE:          3810.5MB/s       3846.8MB/s      6426.8MB/s
  READ:           1956.1MB/s       1994.6MB/s      2145.2MB/s
  WRITE:          1955.9MB/s       1993.5MB/s      2144.8MB/s
  READ:           1852.8MB/s       1911.6MB/s      2075.8MB/s
  WRITE:          1855.7MB/s       1914.6MB/s      2078.1MB/s

perf stat

                                  4 streams                       8 streams                       per-cpu
  ====================================================================================================================
  jobs1
  stalled-cycles-frontend      23,174,811,209 (  38.21%)     23,220,254,188 (  38.25%)       23,061,406,918 (  38.34%)
  stalled-cycles-backend       11,514,174,638 (  18.98%)     11,696,722,657 (  19.27%)       11,370,852,810 (  18.90%)
  instructions                 73,925,005,782 (    1.22)     73,903,177,632 (    1.22)       73,507,201,037 (    1.22)
  branches                     14,455,124,835 ( 756.063)     14,455,184,779 ( 755.281)       14,378,599,509 ( 758.546)
  branch-misses                    69,801,336 (   0.48%)         80,225,529 (   0.55%)           72,044,726 (   0.50%)
  jobs2
  stalled-cycles-frontend      49,912,741,782 (  46.11%)     50,101,189,290 (  45.95%)       32,874,195,633 (  35.11%)
  stalled-cycles-backend       27,080,366,230 (  25.02%)     27,949,970,232 (  25.63%)       16,461,222,706 (  17.58%)
  instructions                122,831,629,690 (    1.13)    122,919,846,419 (    1.13)      121,924,786,775 (    1.30)
  branches                     23,725,889,239 ( 692.663)     23,733,547,140 ( 688.062)       23,553,950,311 ( 794.794)
  branch-misses                    90,733,041 (   0.38%)         96,320,895 (   0.41%)           84,561,092 (   0.36%)
  jobs3
  stalled-cycles-frontend      66,437,834,608 (  45.58%)     63,534,923,344 (  43.69%)       42,101,478,505 (  33.19%)
  stalled-cycles-backend       34,940,799,661 (  23.97%)     34,774,043,148 (  23.91%)       21,163,324,388 (  16.68%)
  instructions                171,692,121,862 (    1.18)    171,775,373,044 (    1.18)      170,353,542,261 (    1.34)
  branches                     32,968,962,622 ( 628.723)     32,987,739,894 ( 630.512)       32,729,463,918 ( 717.027)
  branch-misses                   111,522,732 (   0.34%)        110,472,894 (   0.33%)           99,791,291 (   0.30%)
  jobs4
  stalled-cycles-frontend      98,741,701,675 (  49.72%)     94,797,349,965 (  47.59%)       54,535,655,381 (  33.53%)
  stalled-cycles-backend       54,642,609,615 (  27.51%)     55,233,554,408 (  27.73%)       27,882,323,541 (  17.14%)
  instructions                220,884,807,851 (    1.11)    220,930,887,273 (    1.11)      218,926,845,851 (    1.35)
  branches                     42,354,518,180 ( 592.105)     42,362,770,587 ( 590.452)       41,955,552,870 ( 716.154)
  branch-misses                   138,093,449 (   0.33%)        131,295,286 (   0.31%)          121,794,771 (   0.29%)
  jobs5
  stalled-cycles-frontend     116,219,747,212 (  48.14%)    110,310,397,012 (  46.29%)       66,373,082,723 (  33.70%)
  stalled-cycles-backend       66,325,434,776 (  27.48%)     64,157,087,914 (  26.92%)       32,999,097,299 (  16.76%)
  instructions                270,615,008,466 (    1.12)    270,546,409,525 (    1.14)      268,439,910,948 (    1.36)
  branches                     51,834,046,557 ( 599.108)     51,811,867,722 ( 608.883)       51,412,576,077 ( 729.213)
  branch-misses                   158,197,086 (   0.31%)        142,639,805 (   0.28%)          133,425,455 (   0.26%)
  jobs6
  stalled-cycles-frontend     138,009,414,492 (  48.23%)    139,063,571,254 (  48.80%)       75,278,568,278 (  32.80%)
  stalled-cycles-backend       79,211,949,650 (  27.68%)     79,077,241,028 (  27.75%)       37,735,797,899 (  16.44%)
  instructions                319,763,993,731 (    1.12)    319,937,782,834 (    1.12)      316,663,600,784 (    1.38)
  branches                     61,219,433,294 ( 595.056)     61,250,355,540 ( 598.215)       60,523,446,617 ( 733.706)
  branch-misses                   169,257,123 (   0.28%)        154,898,028 (   0.25%)          141,180,587 (   0.23%)
  jobs7
  stalled-cycles-frontend     162,974,812,119 (  49.20%)    159,290,061,987 (  48.43%)       88,046,641,169 (  33.21%)
  stalled-cycles-backend       92,223,151,661 (  27.84%)     91,667,904,406 (  27.87%)       44,068,454,971 (  16.62%)
  instructions                369,516,432,430 (    1.12)    369,361,799,063 (    1.12)      365,290,380,661 (    1.38)
  branches                     70,795,673,950 ( 594.220)     70,743,136,124 ( 597.876)       69,803,996,038 ( 732.822)
  branch-misses                   181,708,327 (   0.26%)        165,767,821 (   0.23%)          150,109,797 (   0.22%)
  jobs8
  stalled-cycles-frontend     185,000,017,027 (  49.30%)    182,334,345,473 (  48.37%)       99,980,147,041 (  33.26%)
  stalled-cycles-backend      105,753,516,186 (  28.18%)    107,937,830,322 (  28.63%)       51,404,177,181 (  17.10%)
  instructions                418,153,161,055 (    1.11)    418,308,565,828 (    1.11)      413,653,475,581 (    1.38)
  branches                     80,035,882,398 ( 592.296)     80,063,204,510 ( 589.843)       79,024,105,589 ( 730.530)
  branch-misses                   199,764,528 (   0.25%)        177,936,926 (   0.22%)          160,525,449 (   0.20%)
  jobs9
  stalled-cycles-frontend     210,941,799,094 (  49.63%)    204,714,679,254 (  48.55%)      114,251,113,756 (  33.96%)
  stalled-cycles-backend      122,640,849,067 (  28.85%)    122,188,553,256 (  28.98%)       58,360,041,127 (  17.35%)
  instructions                468,151,025,415 (    1.10)    467,354,869,323 (    1.11)      462,665,165,216 (    1.38)
  branches                     89,657,067,510 ( 585.628)     89,411,550,407 ( 588.990)       88,360,523,943 ( 730.151)
  branch-misses                   218,292,301 (   0.24%)        191,701,247 (   0.21%)          178,535,678 (   0.20%)
  jobs10
  stalled-cycles-frontend     233,595,958,008 (  49.81%)    227,540,615,689 (  49.11%)      160,341,979,938 (  43.07%)
  stalled-cycles-backend      136,153,676,021 (  29.03%)    133,635,240,742 (  28.84%)       65,909,135,465 (  17.70%)
  instructions                517,001,168,497 (    1.10)    516,210,976,158 (    1.11)      511,374,038,613 (    1.37)
  branches                     98,911,641,329 ( 585.796)     98,700,069,712 ( 591.583)       97,646,761,028 ( 728.712)
  branch-misses                   232,341,823 (   0.23%)        199,256,308 (   0.20%)          183,135,268 (   0.19%)

per-cpu streams tend to cause significantly less stalled cycles; execute
less branches and hit less branch-misses.

perf stat reported execution time

                          4 streams        8 streams       per-cpu
  ====================================================================
  jobs1
  seconds elapsed        20.909073870     20.875670495    20.817838540
  jobs2
  seconds elapsed        18.529488399     18.720566469    16.356103108
  jobs3
  seconds elapsed        18.991159531     18.991340812    16.766216066
  jobs4
  seconds elapsed        19.560643828     19.551323547    16.246621715
  jobs5
  seconds elapsed        24.746498464     25.221646740    20.696112444
  jobs6
  seconds elapsed        28.258181828     28.289765505    22.885688857
  jobs7
  seconds elapsed        32.632490241     31.909125381    26.272753738
  jobs8
  seconds elapsed        35.651403851     36.027596308    29.108024711
  jobs9
  seconds elapsed        40.569362365     40.024227989    32.898204012
  jobs10
  seconds elapsed        44.673112304     43.874898137    35.632952191

Please see
  Link: http://marc.info/?l=linux-kernel&m=146166970727530
  Link: http://marc.info/?l=linux-kernel&m=146174716719650
for more test results (under low memory conditions).

Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Suggested-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

3 files changed, 116 insertions, 231 deletions

diff --git a/drivers/block/zram/zcomp.c b/drivers/block/zram/zcomp.c
index 3ef42e563bb5..bc98d5ed5477 100644
--- a/drivers/block/zram/zcomp.c
+++ b/drivers/block/zram/zcomp.c
@@ -13,6 +13,7 @@
 #include <linux/slab.h>
 #include <linux/wait.h>
 #include <linux/sched.h>
+#include <linux/cpu.h>
 
 #include "zcomp.h"
 #include "zcomp_lzo.h"
@@ -20,29 +21,6 @@
 #include "zcomp_lz4.h"
 #endif
 
-/*
- * single zcomp_strm backend
- */
-struct zcomp_strm_single {
-	struct mutex strm_lock;
-	struct zcomp_strm *zstrm;
-};
-
-/*
- * multi zcomp_strm backend
- */
-struct zcomp_strm_multi {
-	/* protect strm list */
-	spinlock_t strm_lock;
-	/* max possible number of zstrm streams */
-	int max_strm;
-	/* number of available zstrm streams */
-	int avail_strm;
-	/* list of available strms */
-	struct list_head idle_strm;
-	wait_queue_head_t strm_wait;
-};
-
 static struct zcomp_backend *backends[] = {
 	&zcomp_lzo,
 #ifdef CONFIG_ZRAM_LZ4_COMPRESS
@@ -93,188 +71,6 @@ static struct zcomp_strm *zcomp_strm_alloc(struct zcomp *comp, gfp_t flags)
 	return zstrm;
 }
 
-/*
- * get idle zcomp_strm or wait until other process release
- * (zcomp_strm_release()) one for us
- */
-static struct zcomp_strm *zcomp_strm_multi_find(struct zcomp *comp)
-{
-	struct zcomp_strm_multi *zs = comp->stream;
-	struct zcomp_strm *zstrm;
-
-	while (1) {
-		spin_lock(&zs->strm_lock);
-		if (!list_empty(&zs->idle_strm)) {
-			zstrm = list_entry(zs->idle_strm.next,
-					struct zcomp_strm, list);
-			list_del(&zstrm->list);
-			spin_unlock(&zs->strm_lock);
-			return zstrm;
-		}
-		/* zstrm streams limit reached, wait for idle stream */
-		if (zs->avail_strm >= zs->max_strm) {
-			spin_unlock(&zs->strm_lock);
-			wait_event(zs->strm_wait, !list_empty(&zs->idle_strm));
-			continue;
-		}
-		/* allocate new zstrm stream */
-		zs->avail_strm++;
-		spin_unlock(&zs->strm_lock);
-		/*
-		 * This function can be called in swapout/fs write path
-		 * so we can't use GFP_FS|IO. And it assumes we already
-		 * have at least one stream in zram initialization so we
-		 * don't do best effort to allocate more stream in here.
-		 * A default stream will work well without further multiple
-		 * streams. That's why we use NORETRY | NOWARN.
-		 */
-		zstrm = zcomp_strm_alloc(comp, GFP_NOIO | __GFP_NORETRY |
-					__GFP_NOWARN);
-		if (!zstrm) {
-			spin_lock(&zs->strm_lock);
-			zs->avail_strm--;
-			spin_unlock(&zs->strm_lock);
-			wait_event(zs->strm_wait, !list_empty(&zs->idle_strm));
-			continue;
-		}
-		break;
-	}
-	return zstrm;
-}
-
-/* add stream back to idle list and wake up waiter or free the stream */
-static void zcomp_strm_multi_release(struct zcomp *comp, struct zcomp_strm *zstrm)
-{
-	struct zcomp_strm_multi *zs = comp->stream;
-
-	spin_lock(&zs->strm_lock);
-	if (zs->avail_strm <= zs->max_strm) {
-		list_add(&zstrm->list, &zs->idle_strm);
-		spin_unlock(&zs->strm_lock);
-		wake_up(&zs->strm_wait);
-		return;
-	}
-
-	zs->avail_strm--;
-	spin_unlock(&zs->strm_lock);
-	zcomp_strm_free(comp, zstrm);
-}
-
-/* change max_strm limit */
-static bool zcomp_strm_multi_set_max_streams(struct zcomp *comp, int num_strm)
-{
-	struct zcomp_strm_multi *zs = comp->stream;
-	struct zcomp_strm *zstrm;
-
-	spin_lock(&zs->strm_lock);
-	zs->max_strm = num_strm;
-	/*
-	 * if user has lowered the limit and there are idle streams,
-	 * immediately free as much streams (and memory) as we can.
-	 */
-	while (zs->avail_strm > num_strm && !list_empty(&zs->idle_strm)) {
-		zstrm = list_entry(zs->idle_strm.next,
-				struct zcomp_strm, list);
-		list_del(&zstrm->list);
-		zcomp_strm_free(comp, zstrm);
-		zs->avail_strm--;
-	}
-	spin_unlock(&zs->strm_lock);
-	return true;
-}
-
-static void zcomp_strm_multi_destroy(struct zcomp *comp)
-{
-	struct zcomp_strm_multi *zs = comp->stream;
-	struct zcomp_strm *zstrm;
-
-	while (!list_empty(&zs->idle_strm)) {
-		zstrm = list_entry(zs->idle_strm.next,
-				struct zcomp_strm, list);
-		list_del(&zstrm->list);
-		zcomp_strm_free(comp, zstrm);
-	}
-	kfree(zs);
-}
-
-static int zcomp_strm_multi_create(struct zcomp *comp, int max_strm)
-{
-	struct zcomp_strm *zstrm;
-	struct zcomp_strm_multi *zs;
-
-	comp->destroy = zcomp_strm_multi_destroy;
-	comp->strm_find = zcomp_strm_multi_find;
-	comp->strm_release = zcomp_strm_multi_release;
-	comp->set_max_streams = zcomp_strm_multi_set_max_streams;
-	zs = kmalloc(sizeof(struct zcomp_strm_multi), GFP_KERNEL);
-	if (!zs)
-		return -ENOMEM;
-
-	comp->stream = zs;
-	spin_lock_init(&zs->strm_lock);
-	INIT_LIST_HEAD(&zs->idle_strm);
-	init_waitqueue_head(&zs->strm_wait);
-	zs->max_strm = max_strm;
-	zs->avail_strm = 1;
-
-	zstrm = zcomp_strm_alloc(comp, GFP_KERNEL);
-	if (!zstrm) {
-		kfree(zs);
-		return -ENOMEM;
-	}
-	list_add(&zstrm->list, &zs->idle_strm);
-	return 0;
-}
-
-static struct zcomp_strm *zcomp_strm_single_find(struct zcomp *comp)
-{
-	struct zcomp_strm_single *zs = comp->stream;
-	mutex_lock(&zs->strm_lock);
-	return zs->zstrm;
-}
-
-static void zcomp_strm_single_release(struct zcomp *comp,
-		struct zcomp_strm *zstrm)
-{
-	struct zcomp_strm_single *zs = comp->stream;
-	mutex_unlock(&zs->strm_lock);
-}
-
-static bool zcomp_strm_single_set_max_streams(struct zcomp *comp, int num_strm)
-{
-	/* zcomp_strm_single support only max_comp_streams == 1 */
-	return false;
-}
-
-static void zcomp_strm_single_destroy(struct zcomp *comp)
-{
-	struct zcomp_strm_single *zs = comp->stream;
-	zcomp_strm_free(comp, zs->zstrm);
-	kfree(zs);
-}
-
-static int zcomp_strm_single_create(struct zcomp *comp)
-{
-	struct zcomp_strm_single *zs;
-
-	comp->destroy = zcomp_strm_single_destroy;
-	comp->strm_find = zcomp_strm_single_find;
-	comp->strm_release = zcomp_strm_single_release;
-	comp->set_max_streams = zcomp_strm_single_set_max_streams;
-	zs = kmalloc(sizeof(struct zcomp_strm_single), GFP_KERNEL);
-	if (!zs)
-		return -ENOMEM;
-
-	comp->stream = zs;
-	mutex_init(&zs->strm_lock);
-	zs->zstrm = zcomp_strm_alloc(comp, GFP_KERNEL);
-	if (!zs->zstrm) {
-		kfree(zs);
-		return -ENOMEM;
-	}
-	return 0;
-}
-
 /* show available compressors */
 ssize_t zcomp_available_show(const char *comp, char *buf)
 {
@@ -301,17 +97,17 @@ bool zcomp_available_algorithm(const char *comp)
 
 bool zcomp_set_max_streams(struct zcomp *comp, int num_strm)
 {
-	return comp->set_max_streams(comp, num_strm);
+	return true;
 }
 
 struct zcomp_strm *zcomp_strm_find(struct zcomp *comp)
 {
-	return comp->strm_find(comp);
+	return *get_cpu_ptr(comp->stream);
 }
 
 void zcomp_strm_release(struct zcomp *comp, struct zcomp_strm *zstrm)
 {
-	comp->strm_release(comp, zstrm);
+	put_cpu_ptr(comp->stream);
 }
 
 int zcomp_compress(struct zcomp *comp, struct zcomp_strm *zstrm,
@@ -327,9 +123,83 @@ int zcomp_decompress(struct zcomp *comp, const unsigned char *src,
 	return comp->backend->decompress(src, src_len, dst);
 }
 
+static int __zcomp_cpu_notifier(struct zcomp *comp,
+		unsigned long action, unsigned long cpu)
+{
+	struct zcomp_strm *zstrm;
+
+	switch (action) {
+	case CPU_UP_PREPARE:
+		if (WARN_ON(*per_cpu_ptr(comp->stream, cpu)))
+			break;
+		zstrm = zcomp_strm_alloc(comp, GFP_KERNEL);
+		if (IS_ERR_OR_NULL(zstrm)) {
+			pr_err("Can't allocate a compression stream\n");
+			return NOTIFY_BAD;
+		}
+		*per_cpu_ptr(comp->stream, cpu) = zstrm;
+		break;
+	case CPU_DEAD:
+	case CPU_UP_CANCELED:
+		zstrm = *per_cpu_ptr(comp->stream, cpu);
+		if (!IS_ERR_OR_NULL(zstrm))
+			zcomp_strm_free(comp, zstrm);
+		*per_cpu_ptr(comp->stream, cpu) = NULL;
+		break;
+	default:
+		break;
+	}
+	return NOTIFY_OK;
+}
+
+static int zcomp_cpu_notifier(struct notifier_block *nb,
+		unsigned long action, void *pcpu)
+{
+	unsigned long cpu = (unsigned long)pcpu;
+	struct zcomp *comp = container_of(nb, typeof(*comp), notifier);
+
+	return __zcomp_cpu_notifier(comp, action, cpu);
+}
+
+static int zcomp_init(struct zcomp *comp)
+{
+	unsigned long cpu;
+	int ret;
+
+	comp->notifier.notifier_call = zcomp_cpu_notifier;
+
+	comp->stream = alloc_percpu(struct zcomp_strm *);
+	if (!comp->stream)
+		return -ENOMEM;
+
+	cpu_notifier_register_begin();
+	for_each_online_cpu(cpu) {
+		ret = __zcomp_cpu_notifier(comp, CPU_UP_PREPARE, cpu);
+		if (ret == NOTIFY_BAD)
+			goto cleanup;
+	}
+	__register_cpu_notifier(&comp->notifier);
+	cpu_notifier_register_done();
+	return 0;
+
+cleanup:
+	for_each_online_cpu(cpu)
+		__zcomp_cpu_notifier(comp, CPU_UP_CANCELED, cpu);
+	cpu_notifier_register_done();
+	return -ENOMEM;
+}
+
 void zcomp_destroy(struct zcomp *comp)
 {
-	comp->destroy(comp);
+	unsigned long cpu;
+
+	cpu_notifier_register_begin();
+	for_each_online_cpu(cpu)
+		__zcomp_cpu_notifier(comp, CPU_UP_CANCELED, cpu);
+	__unregister_cpu_notifier(&comp->notifier);
+	cpu_notifier_register_done();
+
+	free_percpu(comp->stream);
 	kfree(comp);
 }
 
@@ -339,9 +209,9 @@ void zcomp_destroy(struct zcomp *comp)
  * backend pointer or ERR_PTR if things went bad. ERR_PTR(-EINVAL)
  * if requested algorithm is not supported, ERR_PTR(-ENOMEM) in
  * case of allocation error, or any other error potentially
- * returned by functions zcomp_strm_{multi,single}_create.
+ * returned by zcomp_init().
  */
-struct zcomp *zcomp_create(const char *compress, int max_strm)
+struct zcomp *zcomp_create(const char *compress)
 {
 	struct zcomp *comp;
 	struct zcomp_backend *backend;
@@ -356,10 +226,7 @@ struct zcomp *zcomp_create(const char *compress, int max_strm)
 		return ERR_PTR(-ENOMEM);
 
 	comp->backend = backend;
-	if (max_strm > 1)
-		error = zcomp_strm_multi_create(comp, max_strm);
-	else
-		error = zcomp_strm_single_create(comp);
+	error = zcomp_init(comp);
 	if (error) {
 		kfree(comp);
 		return ERR_PTR(error);
diff --git a/drivers/block/zram/zcomp.h b/drivers/block/zram/zcomp.h
index b7d2a4bcae54..ffd88cb747fe 100644
--- a/drivers/block/zram/zcomp.h
+++ b/drivers/block/zram/zcomp.h
@@ -10,8 +10,6 @@
 #ifndef _ZCOMP_H_
 #define _ZCOMP_H_
 
-#include <linux/mutex.h>
-
 struct zcomp_strm {
 	/* compression/decompression buffer */
 	void *buffer;
@@ -21,8 +19,6 @@ struct zcomp_strm {
 	 * working memory)
 	 */
 	void *private;
-	/* used in multi stream backend, protected by backend strm_lock */
-	struct list_head list;
 };
 
 /* static compression backend */
@@ -41,19 +37,15 @@ struct zcomp_backend {
 
 /* dynamic per-device compression frontend */
 struct zcomp {
-	void *stream;
+	struct zcomp_strm * __percpu *stream;
 	struct zcomp_backend *backend;
-
-	struct zcomp_strm *(*strm_find)(struct zcomp *comp);
-	void (*strm_release)(struct zcomp *comp, struct zcomp_strm *zstrm);
-	bool (*set_max_streams)(struct zcomp *comp, int num_strm);
-	void (*destroy)(struct zcomp *comp);
+	struct notifier_block notifier;
 };
 
 ssize_t zcomp_available_show(const char *comp, char *buf);
 bool zcomp_available_algorithm(const char *comp);
 
-struct zcomp *zcomp_create(const char *comp, int max_strm);
+struct zcomp *zcomp_create(const char *comp);
 void zcomp_destroy(struct zcomp *comp);
 
 struct zcomp_strm *zcomp_strm_find(struct zcomp *comp);
diff --git a/drivers/block/zram/zram_drv.c b/drivers/block/zram/zram_drv.c
index b09acdb753ee..f92965c4229b 100644
--- a/drivers/block/zram/zram_drv.c
+++ b/drivers/block/zram/zram_drv.c
@@ -650,7 +650,7 @@ static int zram_bvec_write(struct zram *zram, struct bio_vec *bvec, u32 index,
 {
 	int ret = 0;
 	size_t clen;
-	unsigned long handle;
+	unsigned long handle = 0;
 	struct page *page;
 	unsigned char *user_mem, *cmem, *src, *uncmem = NULL;
 	struct zram_meta *meta = zram->meta;
@@ -673,9 +673,8 @@ static int zram_bvec_write(struct zram *zram, struct bio_vec *bvec, u32 index,
 			goto out;
 	}
 
-	zstrm = zcomp_strm_find(zram->comp);
+compress_again:
 	user_mem = kmap_atomic(page);
-
 	if (is_partial_io(bvec)) {
 		memcpy(uncmem + offset, user_mem + bvec->bv_offset,
 		       bvec->bv_len);
@@ -699,6 +698,7 @@ static int zram_bvec_write(struct zram *zram, struct bio_vec *bvec, u32 index,
 		goto out;
 	}
 
+	zstrm = zcomp_strm_find(zram->comp);
 	ret = zcomp_compress(zram->comp, zstrm, uncmem, &clen);
 	if (!is_partial_io(bvec)) {
 		kunmap_atomic(user_mem);
@@ -710,6 +710,7 @@ static int zram_bvec_write(struct zram *zram, struct bio_vec *bvec, u32 index,
 		pr_err("Compression failed! err=%d\n", ret);
 		goto out;
 	}
+
 	src = zstrm->buffer;
 	if (unlikely(clen > max_zpage_size)) {
 		clen = PAGE_SIZE;
@@ -717,8 +718,33 @@ static int zram_bvec_write(struct zram *zram, struct bio_vec *bvec, u32 index,
 			src = uncmem;
 	}
 
-	handle = zs_malloc(meta->mem_pool, clen, GFP_NOIO | __GFP_HIGHMEM);
+	/*
+	 * handle allocation has 2 paths:
+	 * a) fast path is executed with preemption disabled (for
+	 *  per-cpu streams) and has __GFP_DIRECT_RECLAIM bit clear,
+	 *  since we can't sleep;
+	 * b) slow path enables preemption and attempts to allocate
+	 *  the page with __GFP_DIRECT_RECLAIM bit set. we have to
+	 *  put per-cpu compression stream and, thus, to re-do
+	 *  the compression once handle is allocated.
+	 *
+	 * if we have a 'non-null' handle here then we are coming
+	 * from the slow path and handle has already been allocated.
+	 */
+	if (!handle)
+		handle = zs_malloc(meta->mem_pool, clen,
+				__GFP_KSWAPD_RECLAIM |
+				__GFP_NOWARN |
+				__GFP_HIGHMEM);
 	if (!handle) {
+		zcomp_strm_release(zram->comp, zstrm);
+		zstrm = NULL;
+
+		handle = zs_malloc(meta->mem_pool, clen,
+				GFP_NOIO | __GFP_HIGHMEM);
+		if (handle)
+			goto compress_again;
+
 		pr_err("Error allocating memory for compressed page: %u, size=%zu\n",
 			index, clen);
 		ret = -ENOMEM;
@@ -1038,7 +1064,7 @@ static ssize_t disksize_store(struct device *dev,
 	if (!meta)
 		return -ENOMEM;
 
-	comp = zcomp_create(zram->compressor, zram->max_comp_streams);
+	comp = zcomp_create(zram->compressor);
 	if (IS_ERR(comp)) {
 		pr_err("Cannot initialise %s compressing backend\n",
 				zram->compressor);