Re: writeout stalls in current -git
From: Fengguang Wu
Date: Thu Nov 01 2007 - 22:22:00 EST
On Thu, Nov 01, 2007 at 08:00:10PM +0100, Torsten Kaiser wrote:
> On 11/1/07, Torsten Kaiser <just.for.lkml@xxxxxxxxxxxxxx> wrote:
> > On 11/1/07, Fengguang Wu <wfg@xxxxxxxxxxxxxxxx> wrote:
> > > Thank you. Maybe we can start by the applied debug patch :-)
> >
> > Will applied it and try to recreate this.
>
> Patch applied, used emerge to install a 2.6.24-rc1 kernel.
>
> I had no complete stalls, but three times during the move from tmpfs
> to the main xfs the emerge got noticeable slower. There still was
> writeout happening, but as emerge prints out every file it has written
> during the pause not one file was processed.
>
> vmstat 10:
> procs -----------memory---------- ---swap-- -----io---- -system-- ----cpu----
> r b swpd free buff cache si so bi bo in cs us sy id wa
> 0 1 0 3146424 332 614768 0 0 134 1849 438 2515 3 4 91 2
> 0 0 0 3146644 332 614784 0 0 2 1628 507 646 0 2 85 13
> 0 0 0 3146868 332 614868 0 0 5 2359 527 1076 0 3 97 0
> 1 0 0 3144372 332 616148 0 0 96 2829 607 2666 2 5 92 0
> -> normal writeout
> 0 0 0 3140560 332 618144 0 0 152 2764 633 3308 3 6 91 0
> 0 0 0 3137332 332 619908 0 0 114 1801 588 2858 3 4 93 0
> 0 0 0 3136912 332 620136 0 0 20 827 393 1605 1 2 98 0
> -> first stall
'stall': vmstat's output stalls for some time, or emerge stalls for
the next several vmstat lines?
> 0 0 0 3137088 332 620136 0 0 0 557 339 1437 0 1 99 0
> 0 0 0 3137160 332 620136 0 0 0 642 310 1400 0 1 99 0
> 0 0 0 3136588 332 620172 0 0 6 2972 527 1195 0 3 80 16
> 0 0 0 3136276 332 620348 0 0 10 2668 558 1195 0 3 96 0
> 0 0 0 3135228 332 620424 0 0 8 2712 522 1311 0 4 96 0
> 0 0 0 3131740 332 621524 0 0 75 2935 559 2457 2 5 93 0
> 0 0 0 3128348 332 622972 0 0 85 1470 490 2607 3 4 93 0
> 0 0 0 3129292 332 622972 0 0 0 527 353 1398 0 1 99 0
> -> second longer stall
> 0 0 0 3128520 332 623028 0 0 6 488 249 1390 0 1 99 0
> 0 0 0 3128236 332 623028 0 0 0 482 222 1222 0 1 99 0
> 0 0 0 3128408 332 623028 0 0 0 585 269 1301 0 0 99 0
> 0 0 0 3128532 332 623028 0 0 0 610 262 1278 0 0 99 0
> 0 0 0 3128568 332 623028 0 0 0 636 345 1639 0 1 99 0
> 0 0 0 3129032 332 623040 0 0 1 664 337 1466 0 1 99 0
> 0 0 0 3129484 332 623040 0 0 0 658 300 1508 0 0 100 0
> 0 0 0 3129576 332 623040 0 0 0 562 271 1454 0 1 99 0
> 0 0 0 3129736 332 623040 0 0 0 627 278 1406 0 1 99 0
> 0 0 0 3129368 332 623040 0 0 0 507 274 1301 0 1 99 0
> 0 0 0 3129004 332 623040 0 0 0 444 211 1213 0 0 99 0
> 0 1 0 3127260 332 623040 0 0 0 1036 305 1242 0 1 95 4
> 0 0 0 3126280 332 623128 0 0 7 4241 555 1575 1 5 84 10
> 0 0 0 3124948 332 623232 0 0 6 4194 529 1505 1 4 95 0
> 0 0 0 3125228 332 624168 0 0 58 1966 586 1964 2 4 94 0
> -> emerge resumed to normal speed, without any intervention from my side
> 0 0 0 3120932 332 625904 0 0 112 1546 546 2565 3 4 93 0
> 0 0 0 3118012 332 627568 0 0 128 1542 612 2705 3 4 93 0
Interesting, the 'bo' never falls to zero.
>
> >From syslog:
> first stall:
> [ 575.050000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 47259 > global 610 0 0 wc __ tw 1023 sk 0
> [ 586.350000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 50465 > global 6117 0 0 wc _M tw 967 sk 0
> [ 586.360000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 50408 > global 6117 0 0 wc __ tw 1022 sk 0
> [ 599.900000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 53523 > global 11141 0 0 wc __ tw 1009 sk 0
> [ 635.780000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 59397 > global 12757 124 0 wc __ tw 0 sk 0
> [ 638.470000] mm/page-writeback.c 418 balance_dirty_pages: > emerge(6113) 1536 global 11405 51 0 wc __ tw 0 sk 0
> [ 638.820000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 58373 > global 11276 48 0 wc __ tw -1 sk 0
> [ 641.260000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 57348 > global 10565 100 0 wc __ tw 0 sk 0
> [ 643.980000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 56324 > global 9788 103 0 wc __ tw -1 sk 0
> [ 646.120000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 55299 > global 8912 6 0 wc __ tw 0 sk 0
>
> second stall:
> [ 664.040000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 48117 > global 2864 81 0 wc _M tw -13 sk 0
> [ 664.400000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 47080 > global 1995 137 0 wc _M tw 176 sk 0
> [ 664.510000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 46232 > global 1929 267 0 wc __ tw 880 sk 0
> cron[6927]: (root) CMD (test -x /usr/sbin/run-crons && /usr/sbin/run-crons )
> [ 809.560000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 49422 > global 19166 217 0 wc _M tw 380 sk 0
> [ 811.720000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 48778 > global 17969 407 0 wc _M tw -4 sk 0
> [ 813.880000] mm/page-writeback.c 418 balance_dirty_pages: > emerge(6113) 1537 global 16592 233 0 wc _M tw -1 sk 0
> [ 814.710000] mm/page-writeback.c 418 balance_dirty_pages: find(6931) > 1537 global 16132 179 0 wc __ tw -1 sk 0
> [ 814.720000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 47750 > global 16040 271 0 wc _M tw -1 sk 0
> [ 815.040000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 46725 > global 15403 779 0 wc CM tw 324 sk 0
>
> the third stall happend after the emerge was finished. There still was
> ~120Mb of dirty data, but its writeout got much slower over several
> seconds.
> vmstat 10:
> procs -----------memory---------- ---swap-- -----io---- -system-- ----cpu----
> r b swpd free buff cache si so bi bo in cs us sy id wa
> 1 0 0 3096152 332 630424 0 0 81 1503 640 2771 5 4 91 0
> 0 0 0 3101024 332 631588 0 0 279 473 510 1281 5 2 92 1
> -> stall / slowdown starts
> 0 0 0 3147924 332 632384 0 0 78 626 449 1384 0 1 99 0
> 1 0 0 3147940 332 632384 0 0 0 611 388 1387 0 1 99 0
> 0 1 0 3147576 332 632384 0 0 0 939 449 1432 0 1 99 0
> 0 0 0 3145476 332 632384 0 0 0 3592 644 925 0 4 93 3
> -> writeout resumes full speed
> 0 0 0 3147232 332 632480 0 0 0 3108 678 1053 0 3 97 0
> 0 0 0 3146860 332 632480 0 0 0 2497 677 859 0 3 97 0
> 0 0 0 3146720 332 632480 0 0 0 2433 648 839 0 3 97 0
> 0 0 0 3147844 332 632484 0 0 0 2394 625 889 0 3 97 0
> 0 0 0 3148128 332 632484 0 0 0 2204 671 848 0 2 97 0
>
> from syslog:
> [ 848.070000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 48084 > global 13805 0 0 wc _M tw 1008 sk 0
> [ 848.080000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 48068 > global 13805 0 0 wc __ tw 1020 sk 0
> [ 884.090000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 61811 > global 30297 2 0 wc __ tw 862 sk 0
> [ 921.760000] mm/page-writeback.c 418 balance_dirty_pages: cat(7170) > 1541 global 28113 391 0 wc __ tw -5 sk 0
> -> that cat was probably my watch cat /proc/meminfo
> -> during the stall there where no updates visible there
> [ 922.190000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 76871 > global 27735 0 0 wc __ tw -5 sk 0
> [ 923.550000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 75842 > global 26688 106 0 wc _M tw -1 sk 0
> [ 924.940000] mm/page-writeback.c 655 wb_kupdate: pdflush(285) 74817 > global 25698 195 0 wc _M tw 0 sk 0
>
> Apart from my normal kde desktop (no compiz) and the emerge the system was idle.
Interestingly, no background_writeout() appears, but only
balance_dirty_pages() and wb_kupdate. Obviously wb_kupdate won't
block the process.
> If I see the complete stall again, I will post that too.
Thank you, could you run it with the attached new debug patch?
Fengguang
---
mm/page-writeback.c | 29 +++++++++++++++++++++++++++++
1 file changed, 29 insertions(+)
--- linux-2.6.24-git17.orig/mm/page-writeback.c
+++ linux-2.6.24-git17/mm/page-writeback.c
@@ -98,6 +98,26 @@ EXPORT_SYMBOL(laptop_mode);
/* End of sysctl-exported parameters */
+#define writeback_debug_report(n, wbc) do { \
+ __writeback_debug_report(n, wbc, __FILE__, __LINE__, __FUNCTION__); \
+} while (0)
+
+void __writeback_debug_report(long n, struct writeback_control *wbc,
+ const char *file, int line, const char *func)
+{
+ printk(KERN_DEBUG "%s %d %s: %s(%d) %ld "
+ "global %lu %lu %lu "
+ "wc %c%c tw %ld sk %ld\n",
+ file, line, func,
+ current->comm, current->pid, n,
+ global_page_state(NR_FILE_DIRTY),
+ global_page_state(NR_WRITEBACK),
+ global_page_state(NR_UNSTABLE_NFS),
+ wbc->encountered_congestion ? 'C':'_',
+ wbc->more_io ? 'M':'_',
+ wbc->nr_to_write,
+ wbc->pages_skipped);
+}
static void background_writeout(unsigned long _min_pages);
@@ -395,6 +415,7 @@ static void balance_dirty_pages(struct a
pages_written += write_chunk - wbc.nr_to_write;
get_dirty_limits(&background_thresh, &dirty_thresh,
&bdi_thresh, bdi);
+ writeback_debug_report(pages_written, &wbc);
}
/*
@@ -421,6 +442,7 @@ static void balance_dirty_pages(struct a
break; /* We've done our duty */
congestion_wait(WRITE, HZ/10);
+ writeback_debug_report(-pages_written, &wbc);
}
if (bdi_nr_reclaimable + bdi_nr_writeback < bdi_thresh &&
@@ -515,6 +537,11 @@ void throttle_vm_writeout(gfp_t gfp_mask
global_page_state(NR_WRITEBACK) <= dirty_thresh)
break;
congestion_wait(WRITE, HZ/10);
+ printk(KERN_DEBUG "throttle_vm_writeout: "
+ "congestion_wait on %lu+%lu > %lu\n",
+ global_page_state(NR_UNSTABLE_NFS),
+ global_page_state(NR_WRITEBACK),
+ dirty_thresh);
/*
* The caller might hold locks which can prevent IO completion
@@ -557,6 +584,7 @@ static void background_writeout(unsigned
wbc.pages_skipped = 0;
writeback_inodes(&wbc);
min_pages -= MAX_WRITEBACK_PAGES - wbc.nr_to_write;
+ writeback_debug_report(min_pages, &wbc);
if (wbc.nr_to_write > 0 || wbc.pages_skipped > 0) {
/* Wrote less than expected */
if (wbc.encountered_congestion || wbc.more_io)
@@ -630,6 +658,7 @@ static void wb_kupdate(unsigned long arg
wbc.encountered_congestion = 0;
wbc.nr_to_write = MAX_WRITEBACK_PAGES;
writeback_inodes(&wbc);
+ writeback_debug_report(nr_to_write, &wbc);
if (wbc.nr_to_write > 0) {
if (wbc.encountered_congestion || wbc.more_io)
congestion_wait(WRITE, HZ/10);