Hi, I am running my tests now. I should have some results later today, but more conclusively tomorrow.
Regards, JT On Sat, Jan 21, 2017 at 2:05 AM, Parthasarathy Bhuvaragan < parthasarathy.bhuvara...@ericsson.com> wrote: > Hi Ying, > > Sure, I will wait for John as he seems to be able to trigger the race > conditions i couldn't. > > Moreover, I was thinking of posting this series to "net" instead of > net-next as these are primarily bug-fixes. > > /Partha > > > On 01/20/2017 01:12 PM, Xue, Ying wrote: > >> Thanks, it's very good now. You can add my ack-by flag to the whole >> series. >> >> But if possible, please let John help us verify again. >> >> Regards, >> Ying >> >> -----Original Message----- >> From: Parthasarathy Bhuvaragan [mailto:parthasarathy.bhuvarag >> a...@ericsson.com] >> Sent: Friday, January 20, 2017 3:52 AM >> To: tipc-discussion@lists.sourceforge.net; jon.ma...@ericsson.com; Xue, >> Ying; thompa....@gmail.com >> Subject: [PATCH net-next v7 0/6] topology server fixes for nametable soft >> lockup >> >> In this series, we revert the commit 333f796235a527 ("tipc: fix a race >> condition leading to subscriber refcnt bug") and provide an alternate >> solution to fix the race conditions in commits 2-4. >> >> We have to do this as the above commit introduced a nametbl soft lockup >> at module exit as described by patch#4. >> >> --- >> v7: Following updates in Patch #2: >> Fix incorrect deletion of all prior subscriptions until the specified >> subscription. >> Protect exported tipc_subscrp_report_overlap() with subscription >> refcount. >> Ensure that subscription can be freed correctly at subscription timer >> expiry. The >> earlier patch#2 in v5/v6, had refcount bug which prevents the above. >> This was >> introduced when we skipped get/put refcount in >> tipc_subscrb_subscrp_delete(), but >> instead do get in tipc_subscrp_subscribe() before starting the timer. >> Thus the >> subscription_create() initialized the refcount and >> tipc_subscrp_subscribe steps it >> to 2. At subscription timeout, we perform put only once and we cannot >> compensate for >> this additional refcount safely. >> v6: Address krefcount warning for John Thompson in Patch#3 >> v5: Address Ying's comment in Patch #2 to remove del_timer_sync(). >> v4: Address Ying's comment by introducing subscription refcount. >> >> Parthasarathy Bhuvaragan (6): >> tipc: fix nametbl_lock soft lockup at node/link events >> tipc: add subscription refcount to avoid invalid delete >> tipc: fix connection refcount error >> tipc: fix nametbl_lock soft lockup at module exit >> tipc: ignore requests when the connection state is not CONNECTED >> tipc: fix cleanup at module unload >> >> net/tipc/node.c | 9 +++- >> net/tipc/server.c | 48 +++++++++------------ net/tipc/subscr.c | 124 >> ++++++++++++++++++++++++++++++------------------------ >> net/tipc/subscr.h | 1 + >> 4 files changed, 99 insertions(+), 83 deletions(-) >> >> -- >> 2.1.4 >> >> ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot _______________________________________________ tipc-discussion mailing list tipc-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/tipc-discussion
