On Fri, 2012-03-09 at 21:33 -0800, Nicholas A. Bellinger wrote:
> On Fri, 2012-03-09 at 19:15 +0000, Rustad, Mark D wrote:
> > On Mar 8, 2012, at 5:27 PM, Nicholas A. Bellinger wrote:
> >
> > > This is just a friendly heads up that lio-core.git/master has been
> > > updated to v3.3-rc6 code, which currently includes the WIP
> > > se_sess->se_kref patches and iscsi-target and qla_target/tcm_qla2xxx
> > > conversions.
> > >
> > > There is still some ongoing work in this area for target-core that I'm
> > > aiming for v3.4-rc1, so please let us know if you run into any problems.
> >
> > I just updated and started seeing a lot of list_add errors when running with lots of kernel debugging enabled while running pain against 4 FCoE LUNs. There is also one slab corruption reported as well. Here is a partial log of what I got:
> >
> > [ 98.283834] ------------[ cut here ]------------
> > [ 98.284757] WARNING: at /home/mdrustad/git/lio-core/lib/list_debug.c:30 __list_add+0x8f/0xa0()
> > [ 98.284757] Hardware name: Bochs
> > [ 98.284757] list_add corruption. prev->next should be next (ffff88007a15bc10), but was ffff88007140e788. (prev=ffff88007140e788).
> > [ 98.284757] Modules linked in: tcm_fc fcoe libfcoe libfc scsi_transport_fc scsi_tgt iscsi_target_mod target_core_pscsi target_core_file target_core_iblock target_core_mod ixgbe mdio [last unloaded: scsi_wait_scan]
> > [ 98.284757] Pid: 9, comm: kworker/1:0 Not tainted 3.3.0-rc6-mdrlio-02641-g1b354c0 #32
> > [ 98.284757] Call Trace:
> > [ 98.284757] [<ffffffff8103627f>] warn_slowpath_common+0x7f/0xc0
> > [ 98.284757] [<ffffffff81036376>] warn_slowpath_fmt+0x46/0x50
> > [ 98.284757] [<ffffffff812c3d1f>] __list_add+0x8f/0xa0
> > [ 98.284757] [<ffffffffa004ef1e>] ? target_get_sess_cmd+0x3e/0xb0 [target_core_mod]
> > [ 98.284757] [<ffffffffa004ef37>] target_get_sess_cmd+0x57/0xb0 [target_core_mod]
> > [ 98.284757] [<ffffffffa00503e6>] target_submit_tmr+0x96/0x160 [target_core_mod]
> > [ 98.284757] [<ffffffffa01605f0>] ft_send_work+0x2d0/0x330 [tcm_fc]
> > [ 98.284757] [<ffffffff81058687>] process_one_work+0x1b7/0x6c0
> > [ 98.284757] [<ffffffff81058626>] ? process_one_work+0x156/0x6c0
> > [ 98.284757] [<ffffffff814f7a7f>] ? __schedule+0x3ef/0x980
> > [ 98.284757] [<ffffffffa0160320>] ? ft_recv_seq+0xf0/0xf0 [tcm_fc]
> > [ 98.284757] [<ffffffff81058f68>] worker_thread+0x158/0x340
> > [ 98.284757] [<ffffffff81058e10>] ? rescuer_thread+0x240/0x240
> > [ 98.284757] [<ffffffff8105e50e>] kthread+0xae/0xc0
> > [ 98.284757] [<ffffffff815035b4>] kernel_thread_helper+0x4/0x10
> > [ 98.284757] [<ffffffff814f9bb4>] ? retint_restore_args+0x13/0x13
> > [ 98.284757] [<ffffffff8105e460>] ? __init_kthread_worker+0x70/0x70
> > [ 98.284757] [<ffffffff815035b0>] ? gs_change+0x13/0x13
>
> Hi MDR,
>
> Thanks alot for the heads up here.. I'll enable CONFIG_DEBUG_LIST=y and
> will take look another at the TMR release path in tcm_fc(FCoE) ASAP.
>
Hello again MDR,
So it appears this specific breakage is only present in recent
lio-core.git code and is due to the v3.3-rc6 merge with mainline code
causing se_cmd init/setup to be re-added into ft_send_tm(), when the new
target_submit_tmr() usage is already responsible for doing se_cmd
init/setup.
Please go ahead and update your tree to the following HEAD:
commit cd5d3f36a5c4f436c3a26659a3555ba2f01cd560
Author: Nicholas Bellinger <nab@xxxxxxxxxxxxxxx>
Date: Fri Mar 9 22:23:47 2012 -0800
tcm_fc: Remove bogus se_cmd setup in ft_send_tm with target_submit_tmr
Also, I just double checked target-pending.git/for-next, and things are
still looking as expected:
http://git.kernel.org/?p=linux/kernel/git/nab/target-pending.git;a=commitdiff;h=59dcb5ec47965d8d22428db67cbea33a9ec4f347
Thanks again for reporting!
--nab
--
To unsubscribe from this list: send the line "unsubscribe target-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
