Thanks. The chunk shouldn't be invalidated, because we have a ref on it, so it should be safe. You can try moving the unref above the unlock, if you want, but it shouldn't make a difference. More likely, we have a corner case of bad refcounting. Daniel On 7/1/19 9:56 AM, Rungta, Vandana wrote:
Daniel,
Attached is a diff file with the Mdcache patches I have applied to the V2.7.4. If this does not work, I'll create a branch on github in a new repo.
Patches applied: https://github.com/nfs-ganesha/nfs-ganesha/commit/136df4f262c3f9bc29df456eac... https://github.com/nfs-ganesha/nfs-ganesha/commit/c98ad1e238f5db9db5ab8db8ce... https://github.com/nfs-ganesha/nfs-ganesha/commit/f0d5b8d4f6dcce4597459c3eba... https://github.com/nfs-ganesha/nfs-ganesha/commit/e11cd5b9f8d6ffbb46061e7cd2... https://github.com/nfs-ganesha/nfs-ganesha/commit/5439f63b8a1e00f6ff56a4def5... https://github.com/nfs-ganesha/nfs-ganesha/commit/e306c01685b74970ada2259f2b...
Here are the code snippets around the lines of code in the core, after mdcache_find_keyed_reason gets an error.
Could this happen because after releasing the content_lock another thead acquired the lock and invalidated the chunk before this thread acquired the lock again? Potential solution - Unref the chunk before releasing the read lock since it can no longer be trusted after releasing the lock?
if (!has_write) { /* We will have to re-find this dirent after we * re-acquire the lock. */ look_ck = dirent->ck;
PTHREAD_RWLOCK_unlock(&directory->content_lock); PTHREAD_RWLOCK_wrlock(&directory->content_lock); has_write = true;
/* Dropping the content_lock may have * invalidated some or all of the dirents and/or * chunks in this directory. We need to start * over from this point. look_ck is now correct * if the dirent is still cached, and we haven't * changed next_ck, so it's still correct for * reloading the chunk. */ first_pass = true;
>>>>> Line 3133 mdcache_lru_unref_chunk(chunk); chunk = NULL;
/* Now we need to look for this dirent again. * We haven't updated next_ck for this dirent * yet, so it is the right whence to use for a * repopulation readdir if the chunk is * discarded. */ goto again;
Thanks, Vandana
On 7/1/19, 5:40 AM, "Daniel Gryniewicz" <dang@redhat.com> wrote:
Hi, Rungta
Because of the patches you've added, none of the line numbers match any code I have right now. Can you put a branch on github with all your patches applied, and point me to it?
Thanks, Daniel
On 6/29/19 8:24 PM, Rungta, Vandana wrote: > Daniel, > > Core in mdcache when mdcache_lru_unref_chunk. Ref count is 0. > > I started these tests 10 days ago with all the mdcache patches including > the one moving the unref_chunk in readdir_chunked to inside the lock. > The test continues to readdir over 5 million entries while > writing/reading/deleting content. I have 2 test setups – both using a > Windows client that hit this core. > > Let me know if there is additional information you need from the cores. > > Thanks, > > Vandana > > [Thread debugging using libthread_db enabled] > > Using host libthread_db library "/lib64/libthread_db.so.1". > > Core was generated by `bin/ganesha.nfsd -f etc/ganesha/ganesha.conf -p > var/run/ganesha.pid -F'. > > Program terminated with signal 11, Segmentation fault. > > #00x00007fbd14d39c40 in pthread_mutex_lock () from /lib64/libpthread.so.0 > > Missing separate debuginfos, use: debuginfo-install > sgw-nfs-ganesha-2.0.93.0-1.x86_64 > > (gdb) bt > > #00x00007fbd14d39c40 in pthread_mutex_lock () from /lib64/libpthread.so.0 > > #10x000000000052af4a in _mdcache_lru_unref_chunk (chunk=0x3836d1a0, > func=0x598a00 <__func__.23678> "mdcache_readdir_chunked", line=3133) > > at /src/src/FSAL/Stackable_FSALs/FSAL_MDCACHE/mdcache_lru.c:2058 > > #20x0000000000542d84 in mdcache_readdir_chunked (directory=0x664fa280, > whence=37935356, dir_state=0x7fbd0a3dfaf0, > > cb=0x4323ed <populate_dirent>, attrmask=0, eod_met=0x7fbd0a3dffeb) at > /src/src/FSAL/Stackable_FSALs/FSAL_MDCACHE/mdcache_helpers.c:3133 > > #30x000000000053054c in mdcache_readdir (dir_hdl=0x664fa2b8, > whence=0x7fbd0a3dfad0, dir_state=0x7fbd0a3dfaf0, cb=0x4323ed > <populate_dirent>, > > attrmask=0, eod_met=0x7fbd0a3dffeb) at > /src/src/FSAL/Stackable_FSALs/FSAL_MDCACHE/mdcache_handle.c:559 > > #40x0000000000432d14 in fsal_readdir (directory=0x664fa2b8, > cookie=37935356, nbfound=0x7fbd0a3dffec, eod_met=0x7fbd0a3dffeb, attrmask=0, > > cb=0x491d35 <nfs3_readdir_callback>, opaque=0x7fbd0a3dffa0) at > /src/src/FSAL/fsal_helper.c:1164 > > #50x0000000000491b1d in nfs3_readdir (arg=0xa8fc5d48, req=0xa8fc5640, > res=0xa32fbb30) at /src/src/Protocols/NFS/nfs3_readdir.c:289 > > #60x0000000000457c16 in nfs_rpc_process_request (reqdata=0xa8fc5640) at > /src/src/MainNFSD/nfs_worker_thread.c:1328 > > #70x00000000004583d5 in nfs_rpc_valid_NFS (req=0xa8fc5640) at > /src/src/MainNFSD/nfs_worker_thread.c:1548 > > #80x00007fbd15d14034 in svc_vc_decode (req=0xa8fc5640) at > /src/src/libntirpc/src/svc_vc.c:829 > > #90x000000000044adc5 in nfs_rpc_decode_request (xprt=0x7ae54c90, > xdrs=0x36434390) at /src/src/MainNFSD/nfs_rpc_dispatcher_thread.c:1345 > > #10 0x00007fbd15d13f45 in svc_vc_recv (xprt=0x7ae54c90) at > /src/src/libntirpc/src/svc_vc.c:802 > > #11 0x00007fbd15d10689 in svc_rqst_xprt_task (wpe=0x7ae54ea8) at > /src/src/libntirpc/src/svc_rqst.c:769 > > #12 0x00007fbd15d10ae6 in svc_rqst_epoll_events (sr_rec=0x1acd4d0, > n_events=1) at /src/src/libntirpc/src/svc_rqst.c:941 > > #13 0x00007fbd15d10d7b in svc_rqst_epoll_loop (sr_rec=0x1acd4d0) at > /src/src/libntirpc/src/svc_rqst.c:1014 > > #14 0x00007fbd15d10e2e in svc_rqst_run_task (wpe=0x1acd4d0) at > /src/src/libntirpc/src/svc_rqst.c:1050 > > #15 0x00007fbd15d197f6 in work_pool_thread (arg=0x75491e70) at > /src/src/libntirpc/src/work_pool.c:181 > > #16 0x00007fbd14d37de5 in start_thread () from /lib64/libpthread.so.0 > > #17 0x00007fbd1463ef1d in clone () from /lib64/libc.so.6 > > (gdb) select-frame 1 > > (gdb) info locals > > rc = 32701 > > refcnt = 171833376 > > lane = 268435456 > > qlane = 0xe007e24e0 > > __func__ = "_mdcache_lru_unref_chunk" > > (gdb) info args > > chunk = 0x3836d1a0 > > func = 0x598a00 <__func__.23678> "mdcache_readdir_chunked" > > line = 3133 > > (gdb) print *chunk > > $1 = {chunks = {next = 0xed43cd15, prev = 0xa386010002000000}, dirents = > {next = 0x300000003000000, prev = 0x1c00000001000000}, > > parent = 0x7000000aaaaaaaa, chunk_lru = {q = {next = 0x6e776f6e6b6e75, > prev = 0xfefffffffeffffff}, qid = LRU_ENTRY_NONE, refcnt = 0, > > flags = 0, lane = 268435456, cf = 16777283}, reload_ck = 1099511627778, > next_ck = 7236828793636126720, num_entries = 841903471} > > (gdb) select-frame 2 > > (gdb) info locals > > status = {major = ERR_FSAL_NOENT, minor = 0} > > cb_result = DIR_CONTINUE > > entry = 0x0 > > attrs = {request_mask = 0, valid_mask = 1433550, supported = 1433582, > type = REGULAR_FILE, filesize = 51682, fsid = {major = 0, minor = 0}, > > acl = 0x0, fileid = 37935372, mode = 493, numlinks = 1, owner = > 4294967294, group = 4294967294, rawdev = {major = 0, minor = 0}, atime = { > > tv_sec = 1561811134795000, tv_nsec = 0}, creation = {tv_sec = 0, tv_nsec > = 0}, ctime = {tv_sec = 1561811134795000, tv_nsec = 0}, mtime = { > > tv_sec = 1561795757, tv_nsec = 842000000}, chgtime = {tv_sec = > 1561811134795000, tv_nsec = 0}, spaceused = 51682, > > change = 1561811134795000000, generation = 0, expire_time_attr = 60, > fs_locations = 0x0, sec_label = {slai_lfs = {lfs_lfs = 0, lfs_pi = 0}, > > slai_data = {slai_data_len = 0, slai_data_val = 0x0}}} > > dirent = 0xa57f3030 > > has_write = true > > set_first_ck = false > > next_ck = 37935375 > > look_ck = 37935377 > > chunk = 0x3836d1a0 > > first_pass = true > > eod = false > > reload_chunk = false > > __func__ = "mdcache_readdir_chunked" > > __PRETTY_FUNCTION__ = "mdcache_readdir_chunked" > > (gdb) info args > > directory = 0x664fa280 > > whence = 37935356 > > dir_state = 0x7fbd0a3dfaf0 > > cb = 0x4323ed <populate_dirent> > > attrmask = 0 > > eod_met = 0x7fbd0a3dffeb > > (gdb) >