On 03/03/2015 01:55 AM, Jeff Layton wrote:
> On Mon, 2 Mar 2015 15:25:12 +0100
> Daniel Wagner <daniel.wagner@xxxxxxxxxxxx> wrote:
>
>> The locks_insert/delete_block() functions are used for flock, posix
>> and leases types. blocked_lock_lock is used to serialize all access to
>> fl_link, fl_block, fl_next and blocked_hash. Here, we prepare the
>> stage for using blocked_lock_lock to protect blocked_hash.
>>
>> Signed-off-by: Daniel Wagner <daniel.wagner@xxxxxxxxxxxx>
>> Cc: Jeff Layton <jlayton@xxxxxxxxxxxxxxx>
>> Cc: "J. Bruce Fields" <bfields@xxxxxxxxxxxx>
>> Cc: Alexander Viro <viro@xxxxxxxxxxxxxxxxxx>
>> ---
>> fs/locks.c | 48 ++++++++++++++++++++++++++++++++++++++++--------
>> 1 file changed, 40 insertions(+), 8 deletions(-)
>>
>> diff --git a/fs/locks.c b/fs/locks.c
>> index 4498da0..02821dd 100644
>> --- a/fs/locks.c
>> +++ b/fs/locks.c
>> @@ -611,11 +611,20 @@ static void locks_delete_global_blocked(struct file_lock *waiter)
>> */
>> static void __locks_delete_block(struct file_lock *waiter)
>> {
>> - locks_delete_global_blocked(waiter);
>> list_del_init(&waiter->fl_block);
>> waiter->fl_next = NULL;
>> }
>>
>> +/* Posix block variant of __locks_delete_block.
>> + *
>> + * Must be called with blocked_lock_lock held.
>> + */
>> +static void __locks_delete_posix_block(struct file_lock *waiter)
>> +{
>> + locks_delete_global_blocked(waiter);
>> + __locks_delete_block(waiter);
>> +}
>> +
>> static void locks_delete_block(struct file_lock *waiter)
>> {
>> spin_lock(&blocked_lock_lock);
>> @@ -623,6 +632,13 @@ static void locks_delete_block(struct file_lock *waiter)
>> spin_unlock(&blocked_lock_lock);
>> }
>>
>> +static void locks_delete_posix_block(struct file_lock *waiter)
>> +{
>> + spin_lock(&blocked_lock_lock);
>> + __locks_delete_posix_block(waiter);
>> + spin_unlock(&blocked_lock_lock);
>> +}
>> +
>> /* Insert waiter into blocker's block list.
>> * We use a circular list so that processes can be easily woken up in
>> * the order they blocked. The documentation doesn't require this but
>> @@ -639,7 +655,17 @@ static void __locks_insert_block(struct file_lock *blocker,
>> BUG_ON(!list_empty(&waiter->fl_block));
>> waiter->fl_next = blocker;
>> list_add_tail(&waiter->fl_block, &blocker->fl_block);
>> - if (IS_POSIX(blocker) && !IS_OFDLCK(blocker))
>> +}
>> +
>> +/* Posix block variant of __locks_insert_block.
>> + *
>> + * Must be called with flc_lock and blocked_lock_lock held.
>> + */
>> +static void __locks_insert_posix_block(struct file_lock *blocker,
>> + struct file_lock *waiter)
>> +{
>> + __locks_insert_block(blocker, waiter);
>> + if (!IS_OFDLCK(blocker))
>> locks_insert_global_blocked(waiter);
>> }
>>
>
> In many ways OFD locks act more like flock locks than POSIX ones. In
> particular, there is no deadlock detection there, so once your
> conversion is done to more widely use the percpu locks, then you should
> be able to avoid taking the blocked_lock_lock for OFD locks as well.
> The 4th patch in this series doesn't currently do that.
>
> You may want to revisit this patch such that the IS_OFDLCK checks are
> done earlier, so that you can avoid taking the blocked_lock_lock if
> IS_POSIX and !IS_OFDLCK.
Thanks for the explanation. I was not entirely sure what to do here
and forgot to ask.
I have fixed that stuff and now I am testing it. Though it seems
that there is a memory leak which can be triggered with
while true; rm -rf /tmp/a; ./lease02 /tmp/a; done
and this happens also without any of my patches. Still trying to
figure out what's happening. Hopefully I just see a ghost.
slabtop tells me that ftrace_event_field is constantly growing:
Active / Total Objects (% used) : 968819303 / 968828665 (100.0%)
Active / Total Slabs (% used) : 11404623 / 11404623 (100.0%)
Active / Total Caches (% used) : 72 / 99 (72.7%)
Active / Total Size (% used) : 45616199.68K / 45619608.73K (100.0%)
Minimum / Average / Maximum Object : 0.01K / 0.05K / 16.00K
OBJS ACTIVE USE OBJ SIZE SLABS OBJ/SLAB CACHE SIZE NAME
967510630 967510630 2% 0.05K 11382478 85 45529912K ftrace_event_field
154368 154368 100% 0.03K 1206 128 4824K kmalloc-32
121856 121856 100% 0.01K 238 512 952K kmalloc-8
121227 121095 99% 0.08K 2377 51 9508K Acpi-State
This is on proper hardware. On a kvm guest, fasync_cache grows fast and finally the
guest runs out of memory. systemd tries hard to restart everything and fails constantly:
[ 187.021758] systemd invoked oom-killer: gfp_mask=0x200da, order=0, oom_score_adj=0
[ 187.022337] systemd cpuset=/ mems_allowed=0
[ 187.022662] CPU: 3 PID: 1 Comm: systemd Not tainted 4.0.0-rc1+ #380
[ 187.023117] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.7.5-20140709_153950- 04/01/2014
[ 187.023801] ffff88007c918000 ffff88007c9179c8 ffffffff81b4f9be ffffffff8116a9cc
[ 187.024373] 0000000000000000 ffff88007c917a88 ffffffff8116a9d1 000000007c917a58
[ 187.024940] ffffffff8224bc98 ffff88007c917a28 0000000000000092 ffffffff81c1b780
[ 187.025515] Call Trace:
[ 187.025698] [<ffffffff81b4f9be>] dump_stack+0x4c/0x65
[ 187.026083] [<ffffffff8116a9cc>] ? dump_header.isra.13+0x7c/0x450
[ 187.026525] [<ffffffff8116a9d1>] dump_header.isra.13+0x81/0x450
[ 187.026958] [<ffffffff810a45c6>] ? trace_hardirqs_on_caller+0x16/0x240
[ 187.027437] [<ffffffff810a47fd>] ? trace_hardirqs_on+0xd/0x10
[ 187.027859] [<ffffffff814fe5c4>] ? ___ratelimit+0x84/0x110
[ 187.028264] [<ffffffff8116b378>] oom_kill_process+0x1e8/0x4c0
[ 187.028683] [<ffffffff8105fda5>] ? has_ns_capability_noaudit+0x5/0x170
[ 187.029167] [<ffffffff8116baf4>] __out_of_memory+0x4a4/0x510
[ 187.029579] [<ffffffff8116bd2b>] out_of_memory+0x5b/0x80
[ 187.029970] [<ffffffff81170f2e>] __alloc_pages_nodemask+0xa0e/0xb60
[ 187.030434] [<ffffffff811ad863>] read_swap_cache_async+0xe3/0x180
[ 187.030881] [<ffffffff811ad9ed>] swapin_readahead+0xed/0x190
[ 187.031300] [<ffffffff8119bcae>] handle_mm_fault+0xbbe/0x1180
[ 187.031719] [<ffffffff81046bed>] __do_page_fault+0x1ed/0x4c0
[ 187.032138] [<ffffffff81046ecc>] do_page_fault+0xc/0x10
[ 187.032520] [<ffffffff81b5ddc2>] page_fault+0x22/0x30
[ 187.032889] Mem-Info:
[ 187.033066] DMA per-cpu:
[ 187.033254] CPU 0: hi: 0, btch: 1 usd: 0
[ 187.033596] CPU 1: hi: 0, btch: 1 usd: 0
[ 187.033941] CPU 2: hi: 0, btch: 1 usd: 0
[ 187.034292] CPU 3: hi: 0, btch: 1 usd: 0
[ 187.034637] DMA32 per-cpu:
[ 187.034837] CPU 0: hi: 186, btch: 31 usd: 51
[ 187.035185] CPU 1: hi: 186, btch: 31 usd: 0
[ 187.035529] CPU 2: hi: 186, btch: 31 usd: 0
[ 187.035873] CPU 3: hi: 186, btch: 31 usd: 32
[ 187.036221] active_anon:5 inactive_anon:0 isolated_anon:0
[ 187.036221] active_file:238 inactive_file:194 isolated_file:0
[ 187.036221] unevictable:0 dirty:0 writeback:8 unstable:0
[ 187.036221] free:3361 slab_reclaimable:4651 slab_unreclaimable:493909
[ 187.036221] mapped:347 shmem:0 pagetables:400 bounce:0
[ 187.036221] free_cma:0
[ 187.038385] DMA free:7848kB min:44kB low:52kB high:64kB active_anon:4kB inactive_anon:12kB active_file:0kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15992kB managed:15908kB mlocked:0kB dirty:0kB writeback:8kB mapped:4kB shmem:0kB slab_reclaimable:12kB slab_unreclaimable:7880kB kernel_stack:32kB pagetables:36kB unstable:0kB bounce:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:132 all_unreclaimable? yes
[ 187.041138] lowmem_reserve[]: 0 1952 1952 1952
[ 187.041510] DMA32 free:5596kB min:5628kB low:7032kB high:8440kB active_anon:16kB inactive_anon:0kB active_file:952kB inactive_file:772kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2080640kB managed:2004912kB mlocked:0kB dirty:0kB writeback:24kB mapped:1384kB shmem:0kB slab_reclaimable:18592kB slab_unreclaimable:1967756kB kernel_stack:1968kB pagetables:1564kB unstable:0kB bounce:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:12716 all_unreclaimable? yes
[ 187.044442] lowmem_reserve[]: 0 0 0 0
[ 187.044756] DMA: 4*4kB (UEM) 2*8kB (UM) 5*16kB (UEM) 2*32kB (UE) 2*64kB (EM) 3*128kB (UEM) 2*256kB (EM) 3*512kB (UEM) 3*1024kB (UEM) 1*2048kB (R) 0*4096kB = 7856kB
[ 187.046022] DMA32: 190*4kB (UER) 6*8kB (R) 1*16kB (R) 1*32kB (R) 0*64kB 0*128kB 1*256kB (R) 1*512kB (R) 0*1024kB 0*2048kB 1*4096kB (R) = 5720kB
[ 187.047128] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
[ 187.047724] 554 total pagecache pages
[ 187.047991] 60 pages in swap cache
[ 187.048259] Swap cache stats: add 102769, delete 102709, find 75688/136761
[ 187.048748] Free swap = 1041456kB
[ 187.048995] Total swap = 1048572kB
[ 187.049250] 524158 pages RAM
[ 187.049463] 0 pages HighMem/MovableOnly
[ 187.049739] 18953 pages reserved
[ 187.049974] [ pid ] uid tgid total_vm rss nr_ptes nr_pmds swapents oom_score_adj name
[ 187.050587] [ 1293] 0 1293 10283 1 23 2 131 -1000 systemd-udevd
[ 187.051253] [ 1660] 0 1660 12793 57 24 2 134 -1000 auditd
[ 187.051872] [ 1681] 81 1681 6637 1 18 2 124 -900 dbus-daemon
[ 187.052529] [ 1725] 0 1725 20707 0 42 3 216 -1000 sshd
[ 187.053146] [ 2344] 0 2344 3257 0 11 2 49 0 systemd-cgroups
[ 187.053820] [ 2345] 0 2345 3257 0 11 2 55 0 systemd-cgroups
[ 187.054497] [ 2350] 0 2350 3257 0 11 2 35 0 systemd-cgroups
[ 187.055175] [ 2352] 0 2352 3257 0 12 2 37 0 systemd-cgroups
[ 187.055846] [ 2354] 0 2354 3257 0 11 2 43 0 systemd-cgroups
[ 187.056530] [ 2355] 0 2355 3257 0 11 2 40 0 systemd-cgroups
[ 187.057212] [ 2356] 0 2356 3257 0 11 2 44 0 systemd-cgroups
[ 187.057886] [ 2362] 0 2362 3257 0 11 3 33 0 systemd-cgroups
[ 187.058564] [ 2371] 0 2371 3257 0 11 2 33 0 systemd-cgroups
[ 187.059244] [ 2372] 0 2372 3257 0 10 2 44 0 systemd-cgroups
[ 187.059917] [ 2373] 0 2373 3257 0 11 2 39 0 systemd-cgroups
[ 187.060600] [ 2376] 0 2376 3257 0 11 2 34 0 systemd-cgroups
[ 187.061280] [ 2377] 0 2377 3257 0 10 2 43 0 systemd-cgroups
[ 187.061942] [ 2378] 0 2378 3257 0 12 3 34 0 systemd-cgroups
[ 187.062598] [ 2379] 0 2379 27502 0 10 3 33 0 agetty
[ 187.063200] [ 2385] 0 2385 3257 0 12 2 44 0 systemd-cgroups
[ 187.063859] [ 2390] 0 2390 3257 0 11 2 43 0 systemd-cgroups
[ 187.064520] [ 2394] 0 2394 3257 0 11 2 41 0 systemd-cgroups
[ 187.065182] [ 2397] 0 2397 3257 0 11 2 43 0 systemd-cgroups
[ 187.065833] [ 2402] 0 2402 3257 0 11 2 42 0 systemd-cgroups
[ 187.066490] [ 2403] 0 2403 3257 0 11 2 44 0 systemd-cgroups
[ 187.067148] [ 2404] 0 2404 27502 0 13 3 30 0 agetty
[ 187.067743] [ 2410] 0 2410 3257 0 11 2 43 0 systemd-cgroups
[ 187.068407] [ 2413] 0 2413 3257 0 11 2 36 0 systemd-cgroups
[ 187.069072] [ 2416] 0 2416 3257 0 11 2 49 0 systemd-cgroups
[ 187.069720] [ 2417] 0 2417 11861 173 26 2 334 0 (journald)
[ 187.070368] Out of memory: Kill process 2417 ((journald)) score 0 or sacrifice child
[ 187.070943] Killed process 2417 ((journald)) total-vm:47444kB, anon-rss:0kB, file-rss:692kB
[ 187.513857] systemd[1]: Unit systemd-logind.service entered failed state.
[ 188.262477] systemd[1]: Unit systemd-journald.service entered failed state.
[ 188.315222] systemd[1]: systemd-logind.service holdoff time over, scheduling restart.
[ 188.334194] systemd[1]: Stopping Login Service...
[ 188.341556] systemd[1]: Starting Login Service...
[ 188.408787] systemd[1]: systemd-journald.service holdoff time over, scheduling restart.
[ 189.284506] systemd[1]: Stopping Journal Service...
[ 189.330806] systemd[1]: Starting Journal Service...
[ 189.384800] systemd[1]: Started Journal Service.
cheers,
daniel
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
