From 704503d836042d4a4c7685b7036e7de0418fbc0f Mon Sep 17 00:00:00 2001 From: Alexey Dobriyan Date: Tue, 31 Mar 2009 15:23:18 -0700 Subject: mm: fix proc_dointvec_userhz_jiffies "breakage" Addresses http://bugzilla.kernel.org/show_bug.cgi?id=9838 On i386, HZ=1000, jiffies_to_clock_t() converts time in a somewhat strange way from the user's point of view: # echo 500 >/proc/sys/vm/dirty_writeback_centisecs # cat /proc/sys/vm/dirty_writeback_centisecs 499 So, we have 5000 jiffies converted to only 499 clock ticks and reported back. TICK_NSEC = 999848 ACTHZ = 256039 Keeping in-kernel variable in units passed from userspace will fix issue of course, but this probably won't be right for every sysctl. [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: Alexey Dobriyan Cc: Peter Zijlstra Cc: Nick Piggin Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/writeback.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'include/linux/writeback.h') diff --git a/include/linux/writeback.h b/include/linux/writeback.h index 7300ecdc480c..93445477f86a 100644 --- a/include/linux/writeback.h +++ b/include/linux/writeback.h @@ -109,8 +109,8 @@ extern int dirty_background_ratio; extern unsigned long dirty_background_bytes; extern int vm_dirty_ratio; extern unsigned long vm_dirty_bytes; -extern int dirty_writeback_interval; -extern int dirty_expire_interval; +extern unsigned int dirty_writeback_interval; +extern unsigned int dirty_expire_interval; extern int vm_highmem_is_dirtyable; extern int block_dump; extern int laptop_mode; -- cgit v1.2.3 From fafd688e4c0c34da0f3de909881117d374e4c7af Mon Sep 17 00:00:00 2001 From: Peter W Morreale Date: Mon, 6 Apr 2009 19:00:29 -0700 Subject: mm: add /proc controls for pdflush threads Add /proc entries to give the admin the ability to control the minimum and maximum number of pdflush threads. This allows finer control of pdflush on both large and small machines. The rationale is simply one size does not fit all. Admins on large and/or small systems may want to tune the min/max pdflush thread count to best suit their needs. Right now the min/max is hardcoded to 2/8. While probably a fair estimate for smaller machines, large machines with large numbers of CPUs and large numbers of filesystems/block devices may benefit from larger numbers of threads working on different block devices. Even if the background flushing algorithm is radically changed, it is still likely that multiple threads will be involved and admins would still desire finer control on the min/max other than to have to recompile the kernel. The patch adds '/proc/sys/vm/nr_pdflush_threads_min' and '/proc/sys/vm/nr_pdflush_threads_max' with r/w permissions. The minimum value for nr_pdflush_threads_min is 1 and the maximum value is the current value of nr_pdflush_threads_max. This minimum is required since additional thread creation is performed in a pdflush thread itself. The minimum value for nr_pdflush_threads_max is the current value of nr_pdflush_threads_min and the maximum value can be 1000. Documentation/sysctl/vm.txt is also updated. [akpm@linux-foundation.org: fix comment, fix whitespace, use __read_mostly] Signed-off-by: Peter W Morreale Reviewed-by: Rik van Riel Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- Documentation/sysctl/vm.txt | 28 ++++++++++++++++++++++++++++ include/linux/writeback.h | 2 ++ kernel/sysctl.c | 23 +++++++++++++++++++++++ mm/pdflush.c | 31 +++++++++++++++++++------------ 4 files changed, 72 insertions(+), 12 deletions(-) (limited to 'include/linux/writeback.h') diff --git a/Documentation/sysctl/vm.txt b/Documentation/sysctl/vm.txt index 3197fc83bc51..97c4b3284329 100644 --- a/Documentation/sysctl/vm.txt +++ b/Documentation/sysctl/vm.txt @@ -39,6 +39,8 @@ Currently, these files are in /proc/sys/vm: - nr_hugepages - nr_overcommit_hugepages - nr_pdflush_threads +- nr_pdflush_threads_min +- nr_pdflush_threads_max - nr_trim_pages (only if CONFIG_MMU=n) - numa_zonelist_order - oom_dump_tasks @@ -463,6 +465,32 @@ The default value is 0. ============================================================== +nr_pdflush_threads_min + +This value controls the minimum number of pdflush threads. + +At boot time, the kernel will create and maintain 'nr_pdflush_threads_min' +threads for the kernel's lifetime. + +The default value is 2. The minimum value you can specify is 1, and +the maximum value is the current setting of 'nr_pdflush_threads_max'. + +See 'nr_pdflush_threads_max' below for more information. + +============================================================== + +nr_pdflush_threads_max + +This value controls the maximum number of pdflush threads that can be +created. The pdflush algorithm will create a new pdflush thread (up to +this maximum) if no pdflush threads have been available for >= 1 second. + +The default value is 8. The minimum value you can specify is the +current value of 'nr_pdflush_threads_min' and the +maximum is 1000. + +============================================================== + overcommit_memory: This value contains a flag that enables memory overcommitment. diff --git a/include/linux/writeback.h b/include/linux/writeback.h index 93445477f86a..9c1ed1fb6ddb 100644 --- a/include/linux/writeback.h +++ b/include/linux/writeback.h @@ -168,6 +168,8 @@ void writeback_set_ratelimit(void); /* pdflush.c */ extern int nr_pdflush_threads; /* Global so it can be exported to sysctl read-only. */ +extern int nr_pdflush_threads_max; /* Global so it can be exported to sysctl */ +extern int nr_pdflush_threads_min; /* Global so it can be exported to sysctl */ #endif /* WRITEBACK_H */ diff --git a/kernel/sysctl.c b/kernel/sysctl.c index b125e3387568..72eb1a41dcab 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -101,6 +101,7 @@ static int __maybe_unused one = 1; static int __maybe_unused two = 2; static unsigned long one_ul = 1; static int one_hundred = 100; +static int one_thousand = 1000; /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */ static int maxolduid = 65535; @@ -1026,6 +1027,28 @@ static struct ctl_table vm_table[] = { .mode = 0444 /* read-only*/, .proc_handler = &proc_dointvec, }, + { + .ctl_name = CTL_UNNUMBERED, + .procname = "nr_pdflush_threads_min", + .data = &nr_pdflush_threads_min, + .maxlen = sizeof nr_pdflush_threads_min, + .mode = 0644 /* read-write */, + .proc_handler = &proc_dointvec_minmax, + .strategy = &sysctl_intvec, + .extra1 = &one, + .extra2 = &nr_pdflush_threads_max, + }, + { + .ctl_name = CTL_UNNUMBERED, + .procname = "nr_pdflush_threads_max", + .data = &nr_pdflush_threads_max, + .maxlen = sizeof nr_pdflush_threads_max, + .mode = 0644 /* read-write */, + .proc_handler = &proc_dointvec_minmax, + .strategy = &sysctl_intvec, + .extra1 = &nr_pdflush_threads_min, + .extra2 = &one_thousand, + }, { .ctl_name = VM_SWAPPINESS, .procname = "swappiness", diff --git a/mm/pdflush.c b/mm/pdflush.c index 235ac440c44e..f2caf96993f8 100644 --- a/mm/pdflush.c +++ b/mm/pdflush.c @@ -57,6 +57,14 @@ static DEFINE_SPINLOCK(pdflush_lock); */ int nr_pdflush_threads = 0; +/* + * The max/min number of pdflush threads. R/W by sysctl at + * /proc/sys/vm/nr_pdflush_threads_max/min + */ +int nr_pdflush_threads_max __read_mostly = MAX_PDFLUSH_THREADS; +int nr_pdflush_threads_min __read_mostly = MIN_PDFLUSH_THREADS; + + /* * The time at which the pdflush thread pool last went empty */ @@ -68,7 +76,7 @@ static unsigned long last_empty_jifs; * Thread pool management algorithm: * * - The minimum and maximum number of pdflush instances are bound - * by MIN_PDFLUSH_THREADS and MAX_PDFLUSH_THREADS. + * by nr_pdflush_threads_min and nr_pdflush_threads_max. * * - If there have been no idle pdflush instances for 1 second, create * a new one. @@ -134,14 +142,13 @@ static int __pdflush(struct pdflush_work *my_work) * To throttle creation, we reset last_empty_jifs. */ if (time_after(jiffies, last_empty_jifs + 1 * HZ)) { - if (list_empty(&pdflush_list)) { - if (nr_pdflush_threads < MAX_PDFLUSH_THREADS) { - last_empty_jifs = jiffies; - nr_pdflush_threads++; - spin_unlock_irq(&pdflush_lock); - start_one_pdflush_thread(); - spin_lock_irq(&pdflush_lock); - } + if (list_empty(&pdflush_list) && + nr_pdflush_threads < nr_pdflush_threads_max) { + last_empty_jifs = jiffies; + nr_pdflush_threads++; + spin_unlock_irq(&pdflush_lock); + start_one_pdflush_thread(); + spin_lock_irq(&pdflush_lock); } } @@ -153,7 +160,7 @@ static int __pdflush(struct pdflush_work *my_work) */ if (list_empty(&pdflush_list)) continue; - if (nr_pdflush_threads <= MIN_PDFLUSH_THREADS) + if (nr_pdflush_threads <= nr_pdflush_threads_min) continue; pdf = list_entry(pdflush_list.prev, struct pdflush_work, list); if (time_after(jiffies, pdf->when_i_went_to_sleep + 1 * HZ)) { @@ -259,9 +266,9 @@ static int __init pdflush_init(void) * Pre-set nr_pdflush_threads... If we fail to create, * the count will be decremented. */ - nr_pdflush_threads = MIN_PDFLUSH_THREADS; + nr_pdflush_threads = nr_pdflush_threads_min; - for (i = 0; i < MIN_PDFLUSH_THREADS; i++) + for (i = 0; i < nr_pdflush_threads_min; i++) start_one_pdflush_thread(); return 0; } -- cgit v1.2.3 From cd17cbfda004fe5f406c01b318c6378d9895896f Mon Sep 17 00:00:00 2001 From: Jens Axboe Date: Fri, 15 May 2009 11:32:24 +0200 Subject: Revert "mm: add /proc controls for pdflush threads" This reverts commit fafd688e4c0c34da0f3de909881117d374e4c7af. Work is progressing to switch away from pdflush as the process backing for flushing out dirty data. So it seems pointless to add more knobs to control pdflush threads. The original author of the patch did not have any specific use cases for adding the knobs, so we can easily revert this before 2.6.30 to avoid having to maintain this API forever. Signed-off-by: Jens Axboe --- Documentation/sysctl/vm.txt | 28 ---------------------------- include/linux/writeback.h | 2 -- kernel/sysctl.c | 23 ----------------------- mm/pdflush.c | 31 ++++++++++++------------------- 4 files changed, 12 insertions(+), 72 deletions(-) (limited to 'include/linux/writeback.h') diff --git a/Documentation/sysctl/vm.txt b/Documentation/sysctl/vm.txt index b716d33912d8..c302ddf629a0 100644 --- a/Documentation/sysctl/vm.txt +++ b/Documentation/sysctl/vm.txt @@ -39,8 +39,6 @@ Currently, these files are in /proc/sys/vm: - nr_hugepages - nr_overcommit_hugepages - nr_pdflush_threads -- nr_pdflush_threads_min -- nr_pdflush_threads_max - nr_trim_pages (only if CONFIG_MMU=n) - numa_zonelist_order - oom_dump_tasks @@ -469,32 +467,6 @@ The default value is 0. ============================================================== -nr_pdflush_threads_min - -This value controls the minimum number of pdflush threads. - -At boot time, the kernel will create and maintain 'nr_pdflush_threads_min' -threads for the kernel's lifetime. - -The default value is 2. The minimum value you can specify is 1, and -the maximum value is the current setting of 'nr_pdflush_threads_max'. - -See 'nr_pdflush_threads_max' below for more information. - -============================================================== - -nr_pdflush_threads_max - -This value controls the maximum number of pdflush threads that can be -created. The pdflush algorithm will create a new pdflush thread (up to -this maximum) if no pdflush threads have been available for >= 1 second. - -The default value is 8. The minimum value you can specify is the -current value of 'nr_pdflush_threads_min' and the -maximum is 1000. - -============================================================== - overcommit_memory: This value contains a flag that enables memory overcommitment. diff --git a/include/linux/writeback.h b/include/linux/writeback.h index 9c1ed1fb6ddb..93445477f86a 100644 --- a/include/linux/writeback.h +++ b/include/linux/writeback.h @@ -168,8 +168,6 @@ void writeback_set_ratelimit(void); /* pdflush.c */ extern int nr_pdflush_threads; /* Global so it can be exported to sysctl read-only. */ -extern int nr_pdflush_threads_max; /* Global so it can be exported to sysctl */ -extern int nr_pdflush_threads_min; /* Global so it can be exported to sysctl */ #endif /* WRITEBACK_H */ diff --git a/kernel/sysctl.c b/kernel/sysctl.c index ea78fa101ad6..b2970d56fb76 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -101,7 +101,6 @@ static int __maybe_unused one = 1; static int __maybe_unused two = 2; static unsigned long one_ul = 1; static int one_hundred = 100; -static int one_thousand = 1000; /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */ static unsigned long dirty_bytes_min = 2 * PAGE_SIZE; @@ -1033,28 +1032,6 @@ static struct ctl_table vm_table[] = { .mode = 0444 /* read-only*/, .proc_handler = &proc_dointvec, }, - { - .ctl_name = CTL_UNNUMBERED, - .procname = "nr_pdflush_threads_min", - .data = &nr_pdflush_threads_min, - .maxlen = sizeof nr_pdflush_threads_min, - .mode = 0644 /* read-write */, - .proc_handler = &proc_dointvec_minmax, - .strategy = &sysctl_intvec, - .extra1 = &one, - .extra2 = &nr_pdflush_threads_max, - }, - { - .ctl_name = CTL_UNNUMBERED, - .procname = "nr_pdflush_threads_max", - .data = &nr_pdflush_threads_max, - .maxlen = sizeof nr_pdflush_threads_max, - .mode = 0644 /* read-write */, - .proc_handler = &proc_dointvec_minmax, - .strategy = &sysctl_intvec, - .extra1 = &nr_pdflush_threads_min, - .extra2 = &one_thousand, - }, { .ctl_name = VM_SWAPPINESS, .procname = "swappiness", diff --git a/mm/pdflush.c b/mm/pdflush.c index f2caf96993f8..235ac440c44e 100644 --- a/mm/pdflush.c +++ b/mm/pdflush.c @@ -57,14 +57,6 @@ static DEFINE_SPINLOCK(pdflush_lock); */ int nr_pdflush_threads = 0; -/* - * The max/min number of pdflush threads. R/W by sysctl at - * /proc/sys/vm/nr_pdflush_threads_max/min - */ -int nr_pdflush_threads_max __read_mostly = MAX_PDFLUSH_THREADS; -int nr_pdflush_threads_min __read_mostly = MIN_PDFLUSH_THREADS; - - /* * The time at which the pdflush thread pool last went empty */ @@ -76,7 +68,7 @@ static unsigned long last_empty_jifs; * Thread pool management algorithm: * * - The minimum and maximum number of pdflush instances are bound - * by nr_pdflush_threads_min and nr_pdflush_threads_max. + * by MIN_PDFLUSH_THREADS and MAX_PDFLUSH_THREADS. * * - If there have been no idle pdflush instances for 1 second, create * a new one. @@ -142,13 +134,14 @@ static int __pdflush(struct pdflush_work *my_work) * To throttle creation, we reset last_empty_jifs. */ if (time_after(jiffies, last_empty_jifs + 1 * HZ)) { - if (list_empty(&pdflush_list) && - nr_pdflush_threads < nr_pdflush_threads_max) { - last_empty_jifs = jiffies; - nr_pdflush_threads++; - spin_unlock_irq(&pdflush_lock); - start_one_pdflush_thread(); - spin_lock_irq(&pdflush_lock); + if (list_empty(&pdflush_list)) { + if (nr_pdflush_threads < MAX_PDFLUSH_THREADS) { + last_empty_jifs = jiffies; + nr_pdflush_threads++; + spin_unlock_irq(&pdflush_lock); + start_one_pdflush_thread(); + spin_lock_irq(&pdflush_lock); + } } } @@ -160,7 +153,7 @@ static int __pdflush(struct pdflush_work *my_work) */ if (list_empty(&pdflush_list)) continue; - if (nr_pdflush_threads <= nr_pdflush_threads_min) + if (nr_pdflush_threads <= MIN_PDFLUSH_THREADS) continue; pdf = list_entry(pdflush_list.prev, struct pdflush_work, list); if (time_after(jiffies, pdf->when_i_went_to_sleep + 1 * HZ)) { @@ -266,9 +259,9 @@ static int __init pdflush_init(void) * Pre-set nr_pdflush_threads... If we fail to create, * the count will be decremented. */ - nr_pdflush_threads = nr_pdflush_threads_min; + nr_pdflush_threads = MIN_PDFLUSH_THREADS; - for (i = 0; i < nr_pdflush_threads_min; i++) + for (i = 0; i < MIN_PDFLUSH_THREADS; i++) start_one_pdflush_thread(); return 0; } -- cgit v1.2.3