So far no deadlock on the socket lock is seen with this patch. I have seen two out-of-memory crashes which leaves the system deadlocked trying to allocate memory. Stack traces and kernel log from one of the oom crashes is attached. I was thinking this may be caused by having the dirty_ratio set too high given our product's heavy use of memory. Currently re-testing with dirty_ratio reduced to 5%. [11515.530000] Recording maint invoked oom-killer: gfp_mask=0x201da, order=0, oom_score_adj=0 [11515.540000] CPU: 0 PID: 1304 Comm: Recording maint Tainted: G O 3.10.32 #10 [11515.550000] Stack : 80666e52 00000049 80660000 00000000 805563dc 8b1876f8 8054af80 805d2227 00000518 00000000 80656d40 8b1876f8 000007ce 00000002 00000000 8047d858 00000000 8003bb64 00000006 00000000 8054d024 [11515.580000] 8b189a84 8b189a84 8054af80 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 8b189a18 ... [11515.600000] Call Trace: [11515.600000] [<8001d574>] show_stack+0x64/0x7c [11515.610000] [<8047f438>] dump_header.isra.15+0x80/0x234 [11515.610000] [<800dad58>] oom_kill_process+0x2b8/0x464 [11515.620000] [<800db480>] out_of_memory+0x330/0x35c [11515.620000] [<800defb8>] __alloc_pages_nodemask+0x748/0x75c [11515.630000] [<800d9ab0>] filemap_fault+0x1b8/0x4b0 [11515.640000] [<800f8da8>] __do_fault+0x88/0x638 [11515.650000] [<800fc360>] handle_pte_fault+0xc4/0x1328 [11515.650000] [<800fd6ec>] handle_mm_fault+0x128/0x188 [11515.660000] [<800253ac>] do_page_fault+0x10c/0x520 [11515.660000] [<80017ee0>] ret_from_exception+0x0/0x10 [11515.670000] [11515.670000] Mem-Info: [11515.670000] Normal per-cpu: [11515.680000] CPU 0: hi: 90, btch: 15 usd: 12 [11515.680000] CPU 1: hi: 90, btch: 15 usd: 88 [11515.690000] active_anon:8293 inactive_anon:12641 isolated_anon:0 [11515.690000] active_file:9 inactive_file:247 isolated_file:0 [11515.690000] unevictable:0 dirty:0 writeback:0 unstable:0 [11515.690000] free:2034 slab_reclaimable:1371 slab_unreclaimable:20541 [11515.690000] mapped:729 shmem:14591 pagetables:330 bounce:0 [11515.690000] free_cma:0 [11515.730000] Normal free:8136kB min:8192kB low:10240kB high:12288kB active_anon:33172kB inactive_anon:50564kB active_file:36kB inactive_file:1024kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:204800kB managed:190524kB mlocked:0kB dirty:0kB writeback:0kB mapped:2916kB shmem:58364kB slab_reclaimable:5484kB slab_unreclaimable:82164kB kernel_stack:1256kB pagetables:1320kB unstable:0kB bounce:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:340 all_unreclaimable? yes [11515.780000] lowmem_reserve[]: 0 0 [11515.780000] Normal: 32*4kB (UM) 0*8kB 0*16kB 0*32kB 1*64kB (R) 1*128kB (R) 1*256kB (R) 1*512kB (R) 1*1024kB (R) 1*2048kB (R) 1*4096kB (R) = 8256kB [11515.800000] 14940 total pagecache pages [11515.820000] 51200 pages RAM [11515.820000] 3486 pages reserved [11515.830000] 263787 pages shared [11515.830000] 44113 pages non-shared Kernel thread cifsd: -000 |__schedule() -001 |__lock_sock(sk = 0x8BFD7920) -002 |lock_sock_nested(sk = 0x8BFD7920, subclass = 0) -003 |sk_wait_data(sk = 0x8BFD7920, timeo = 0x8AFD3C58) -004 |tcp_recvmsg(?, sk = 0x8BFD7920, msg = 0x8AFD3DE0, len = 4, nonblock = 0, f -005 |inet_recvmsg(iocb = 0x8AFD3CF8, ?, msg = 0x8AFD3DE0, size = 4, flags = 0) -006 |sock_recvmsg(sock = 0x87C0A8E0, msg = 0x8AFD3DE0, size = 4, flags = 0) -007 |kernel_recvmsg(?, ?, ?, ?, size = 4, flags = 0) -008 |cifs_readv_from_socket(server = 0x8B072800, iov_orig = 0x8AFD3E40, nr_segs -009 |cifs_read_from_socket(?, ?, ?) -010 |cifs_demultiplex_thread(p = 0x8B072800) -011 |kthread(_create = 0x8C7DFD40) -012 |ret_from_kernel_thread(asm) --> |exception --- |end of frame Kernel thread kworker/1:1 -000 |__schedule() -001 |__cond_resched() -002 |cond_resched() -003 |shrink_slab(shrink = 0x8C6C799C, nr_pages_scanned = 32, lru_pages = 5) -004 |try_to_free_pages(?, ?, ?, ?) -005 |__alloc_pages_nodemask(?, order = 0, ?, nodemask = 0x0) -006 |cache_alloc_refill(cachep = 0x8C414780, flags = 80, ?) -007 |kmem_cache_alloc(cachep = 0x8C414780, flags = 80) -008 |__alloc_skb(size = 1856, gfp_mask = 80, ?, ?) -009 |sk_stream_alloc_skb(sk = 0x8BFD7920, size = 1648, ?) -010 |tcp_sendmsg(?, sk = 0x8BFD7920, ?, ?) -011 |sock_sendmsg(sock = 0x87C0A8E0, msg = 0x8C6C7CB8, size = 42) -012 |kernel_sendmsg(?, ?, ?, ?, size = 42) -013 |smb_send_kvec(server = 0x8B072800, iov = 0x8C6C7DC8, n_vec = 1, sent = 0x8C6C7D2C) -014 |smb_send_rqst(server = 0x8B072800, rqst = 0x8C6C7DB0) -015 |cifs_call_async(server = 0x8B072800, rqst = 0x8C6C7DB0, receive = 0x0, callback = 0x802194DC, cb -016 |CIFSSMBEcho(server = 0x8B072800) -017 |cifs_echo_request(work = 0x8B072A90) -018 |process_one_work(worker = 0x8C6F5B80, work = 0x8B072A90) -019 |worker_thread(__worker = 0x8C6F5B80) -020 |kthread(_create = 0x8C47DD90) -021 |ret_from_kernel_thread(asm) --> |exception --- |end of frame sk = 0x8BFD7920 -> ( __sk_common = (skc_addrpair = 8575039233117432000, skc_daddr = 16820416, skc_rcv_saddr = 1996531904, skc_hash = 670815510, skc_u16hashes = (54550, 10235), skc_portpair = 3106389249, skc_dport = 48385, skc_num = 47399, skc_f sk_lock = ( slock = ( rlock = ( raw_lock = ( lock = 0x00DC00DC, h = (serving_now = 220, ticket = 220)), magic = 0xDEAD4EAD, owner_cpu = 0xFFFFFFFF, owner = 0xFFFFFFFF, dep_map = (key = 0x80B87FC0, class_cache = (0x0, 0x0), name = 0x80568D0C)), __padding = (220, 0, 220, 0, 173, 78, 173, 222, 255, 255, 255, 255, 255, 255, 255, 255), dep_map = (key = 0x80B87FC0, class_cache = (0x0, 0x0), name = 0x80568D0C)), owned = 1, wq = (lock = (rlock = (raw_lock = (lock = 196611, h = (serving_now = 3, ticket = 3)), magic = 3735899821, owner_cpu = 4294967295, owner = 0xFFFFFFFF, dep_map = (key = 0x80B87FE0, class_cache = (0x0, 0x0), name = 0x80568CB dep_map = (key = 0x80B87FD0, class_cache = (0x0, 0x0), name = 0x80568D20)), > -----Original Message----- > From: Eric Dumazet [mailto:eric.dumazet@xxxxxxxxx] > Sent: den 6 mars 2014 03:21 > To: David Miller > Cc: Lars Persson; netdev@xxxxxxxxxxxxxxx > Subject: Re: [PATCH] tcp: tcp_release_cb() should release socket > ownership > > On Wed, 2014-03-05 at 21:15 -0500, David Miller wrote: > > > Targetting net-next for the assertion is fine. > > > > Did you get test results back yet? > > Lars gave the results this morning for the first patch. > > He said he was now testing the official patch. > > Lets wait a bit more ;) > > ��.n��������+%������w��{.n����z����ܨ}���Ơz�j:+v�����w����ޙ��&�)ߡ�a����z�ޗ���ݢj��w�f