| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Suppose we have an environment with a number of non-NPIV FCP devices
(virtual HBAs / FCP devices / zfcp "adapter"s) sharing the same physical
FCP channel (HBA port) and its I_T nexus. Plus a number of storage target
ports zoned to such shared channel. Now one target port logs out of the
fabric causing an RSCN. Zfcp reacts with an ADISC ELS and subsequent port
recovery depending on the ADISC result. This happens on all such FCP
devices (in different Linux images) concurrently as they all receive a copy
of this RSCN. In the following we look at one of those FCP devices.
Requests other than FSF_QTCB_FCP_CMND can be slow until they get a
response.
Depending on which requests are affected by slow responses, there are
different recovery outcomes. Here we want to fix failed recoveries on port
or adapter level by avoiding recovery requests that can be slow.
We need the cached N_Port_ID for the remote port "link" test with ADISC.
Just before sending the ADISC, we now intentionally forget the old cached
N_Port_ID. The idea is that on receiving an RSCN for a port, we have to
assume that any cached information about this port is stale. This forces a
fresh new GID_PN [FC-GS] nameserver lookup on any subsequent recovery for
the same port. Since we typically can still communicate with the nameserver
efficiently, we now reach steady state quicker: Either the nameserver still
does not know about the port so we stop recovery, or the nameserver already
knows the port potentially with a new N_Port_ID and we can successfully and
quickly perform open port recovery. For the one case, where ADISC returns
successfully, we re-initialize port->d_id because that case does not
involve any port recovery.
This also solves a problem if the storage WWPN quickly logs into the fabric
again but with a different N_Port_ID. Such as on virtual WWPN takeover
during target NPIV failover.
[https://www.redbooks.ibm.com/abstracts/redp5477.html] In that case the
RSCN from the storage FDISC was ignored by zfcp and we could not
successfully recover the failover. On some later failback on the storage,
we could have been lucky if the virtual WWPN got the same old N_Port_ID
from the SAN switch as we still had cached. Then the related RSCN
triggered a successful port reopen recovery. However, there is no
guarantee to get the same N_Port_ID on NPIV FDISC.
Even though NPIV-enabled FCP devices are not affected by this problem, this
code change optimizes recovery time for gone remote ports as a side effect.
The timely drop of cached N_Port_IDs prevents unnecessary slow open port
attempts.
While the problem might have been in code before v2.6.32 commit
799b76d09aee ("[SCSI] zfcp: Decouple gid_pn requests from erp") this fix
depends on the gid_pn_work introduced with that commit, so we mark it as
culprit to satisfy fix dependencies.
Note: Point-to-point remote port is already handled separately and gets its
N_Port_ID from the cached peer_d_id. So resetting port->d_id in general
does not affect PtP.
Link: https://lore.kernel.org/r/20220118165803.3667947-1-maier@linux.ibm.com
Fixes: 799b76d09aee ("[SCSI] zfcp: Decouple gid_pn requests from erp")
Cc: <stable@vger.kernel.org> #2.6.32+
Suggested-by: Benjamin Block <bblock@linux.ibm.com>
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
The callers know what type of queue they want to work with. Introduce
type-specific variants to add buffers on an {Input,Output} queue, so
that we can avoid some function parameters and the de-muxing into
type-specific hot paths.
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
|
|
The callers know what type of queue they want to inspect. Introduce
type-specific variants to inspect an {Input,Output} queue, so that we
can avoid one function parameter and some conditional branches in the
hot paths.
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
|
|
struct device supports attribute groups directly but does not support
struct device_attribute directly. Hence switch to attribute groups.
Link: https://lore.kernel.org/r/20211012233558.4066756-7-bvanassche@acm.org
Acked-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
Conditional statements are faster than indirect calls. Hence call
scsi_done() directly.
Link: https://lore.kernel.org/r/20211007202923.2174984-8-bvanassche@acm.org
Acked-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux
Pull more s390 updates from Heiko Carstens:
"Except for the xpram device driver removal it is all about fixes and
cleanups.
- Fix topology update on cpu hotplug, so notifiers see expected
masks. This bug was uncovered with SCHED_CORE support.
- Fix stack unwinding so that the correct number of entries are
omitted like expected by common code. This fixes KCSAN selftests.
- Add kmemleak annotation to stack_alloc to avoid false positive
kmemleak warnings.
- Avoid layering violation in common I/O code and don't unregister
subchannel from child-drivers.
- Remove xpram device driver for which no real use case exists since
the kernel is 64 bit only. Also all hypervisors got required
support removed in the meantime, which means the xpram device
driver is dead code.
- Fix -ENODEV handling of clp_get_state in our PCI code.
- Enable KFENCE in debug defconfig.
- Cleanup hugetlbfs s390 specific Kconfig dependency.
- Quite a lot of trivial fixes to get rid of "W=1" warnings, and and
other simple cleanups"
* tag 's390-5.15-2' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
hugetlbfs: s390 is always 64bit
s390/ftrace: remove incorrect __va usage
s390/zcrypt: remove incorrect kernel doc indicators
scsi: zfcp: fix kernel doc comments
s390/sclp: add __nonstring annotation
s390/hmcdrv_ftp: fix kernel doc comment
s390: remove xpram device driver
s390/pci: read clp_list_pci_req only once
s390/pci: fix clp_get_state() handling of -ENODEV
s390/cio: fix kernel doc comment
s390/ctrlchar: fix kernel doc comment
s390/con3270: use proper type for tasklet function
s390/cpum_cf: move array from header to C file
s390/mm: fix kernel doc comments
s390/topology: fix topology information when calling cpu hotplug notifiers
s390/unwind: use current_frame_address() to unwind current task
s390/configs: enable CONFIG_KFENCE in debug_defconfig
s390/entry: make oklabel within CHKSTG macro local
s390: add kmemleak annotation in stack_alloc()
s390/cio: dont unregister subchannel from child-drivers
|
|
A couple of function names don't match what the kernel doc comments
indicate.
Acked-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
|
|
Pull SCSI updates from James Bottomley:
"This series consists of the usual driver updates (ufs, qla2xxx,
target, smartpqi, lpfc, mpt3sas).
The core change causing the most churn was replacing the command
request field request with a macro, allowing us to offset map to it
and remove the redundant field; the same was also done for the tag
field.
The most impactful change is the final removal of scsi_ioctl, which
has been deprecated for over a decade"
* tag 'scsi-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi: (293 commits)
scsi: ufs: Fix ufshcd_request_sense_async() for Samsung KLUFG8RHDA-B2D1
scsi: ufs: ufs-exynos: Fix static checker warning
scsi: mpt3sas: Use the proper SCSI midlayer interfaces for PI
scsi: lpfc: Use the proper SCSI midlayer interfaces for PI
scsi: lpfc: Copyright updates for 14.0.0.1 patches
scsi: lpfc: Update lpfc version to 14.0.0.1
scsi: lpfc: Add bsg support for retrieving adapter cmf data
scsi: lpfc: Add cmf_info sysfs entry
scsi: lpfc: Add debugfs support for cm framework buffers
scsi: lpfc: Add support for maintaining the cm statistics buffer
scsi: lpfc: Add rx monitoring statistics
scsi: lpfc: Add support for the CM framework
scsi: lpfc: Add cmfsync WQE support
scsi: lpfc: Add support for cm enablement buffer
scsi: lpfc: Add cm statistics buffer support
scsi: lpfc: Add EDC ELS support
scsi: lpfc: Expand FPIN and RDF receive logging
scsi: lpfc: Add MIB feature enablement support
scsi: lpfc: Add SET_HOST_DATA mbox cmd to pass date/time info to firmware
scsi: fc: Add EDC ELS definition
...
|
|
Prepare for removal of the request pointer by using scsi_cmd_to_rq()
instead. This patch does not change any functionality.
Link: https://lore.kernel.org/r/20210809230355.8186-11-bvanassche@acm.org
Acked-by: Benjamin Block <bblock@linux.ibm.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
Now that all drivers use qdio_inspect_queue() and qdio's internal
queue tasklets are gone, the driver-specified queue handlers are
only called for async error reporting (eg. for an error condition in
the QEBSM code).
So take a moment to clean up the Output Queue handlers (they are
_always_ called with qdio_error != 0), and clarify which error types
can be reported through what interface. As Benjamin already suggested
a while ago, we should turn these into distinct enums at some point.
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
|
|
Use scsi_prot_ref_tag() and scsi_prot_interval() instead scsi_get_lba() and
sector_size.
Link: https://lore.kernel.org/r/20210609033929.3815-7-martin.petersen@oracle.com
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Message-Id: <20210609033929.3815-7-martin.petersen@oracle.com>
|
|
On remote cable pull, a zfcp_port keeps its status and only gets
ZFCP_STATUS_PORT_LINK_TEST added. Only after an ADISC timeout, we would
actually start port recovery and remove ZFCP_STATUS_COMMON_UNBLOCKED which
zfcp_sysfs_port_fc_security_show() detected and reported as "unknown"
instead of the old and possibly stale zfcp_port->connection_info.
Add check for ZFCP_STATUS_PORT_LINK_TEST for timely "unknown" report.
Link: https://lore.kernel.org/r/20210702160922.2667874-1-maier@linux.ibm.com
Fixes: a17c78460093 ("scsi: zfcp: report FC Endpoint Security in sysfs")
Cc: <stable@vger.kernel.org> #5.7+
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
Introduce scsi_build_sense() as a wrapper around scsi_build_sense_buffer()
to format the buffer and set the correct SCSI status.
Link: https://lore.kernel.org/r/20210427083046.31620-8-hare@suse.de
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
Pull SCSI updates from James Bottomley:
"This consists of the usual driver updates (ufs, target, tcmu,
smartpqi, lpfc, zfcp, qla2xxx, mpt3sas, pm80xx).
The major core change is using a sbitmap instead of an atomic for
queue tracking"
* tag 'scsi-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi: (412 commits)
scsi: target: tcm_fc: Fix a kernel-doc header
scsi: target: Shorten ALUA error messages
scsi: target: Fix two format specifiers
scsi: target: Compare explicitly with SAM_STAT_GOOD
scsi: sd: Introduce a new local variable in sd_check_events()
scsi: dc395x: Open-code status_byte(u8) calls
scsi: 53c700: Open-code status_byte(u8) calls
scsi: smartpqi: Remove unused functions
scsi: qla4xxx: Remove an unused function
scsi: myrs: Remove unused functions
scsi: myrb: Remove unused functions
scsi: mpt3sas: Fix two kernel-doc headers
scsi: fcoe: Suppress a compiler warning
scsi: libfc: Fix a format specifier
scsi: aacraid: Remove an unused function
scsi: core: Introduce enum scsi_disposition
scsi: core: Modify the scsi_send_eh_cmnd() return value for the SDEV_BLOCK case
scsi: core: Rename scsi_softirq_done() into scsi_complete()
scsi: core: Remove an incorrect comment
scsi: core: Make the scsi_alloc_sgtables() documentation more accurate
...
|
|
The qdio layer currently provides its own infrastructure to scan for
Request Queue completions & to report them to the device driver. This
comes with several drawbacks - having an async tasklet & timer construct in
qdio introduces additional lifetime complexity, and makes it harder to
integrate them with the rest of the device driver. The timeouts are also
currently hard-coded, and can't be tweaked without affecting other qdio
drivers (ie. qeth).
But due to recent enhancements to the qdio layer, zfcp can actually take
full control of the Request Queue completion processing. It merely needs to
opt-out from the qdio layer mechanisms by setting the scan_threshold to 0,
and then use qdio_inspect_queue() to scan for completions.
So re-implement the tasklet & timer mechanism in zfcp, while initially
copying the scan conditions from qdio's handle_outbound() and
qdio_outbound_tasklet(). One minor behavioural change is that
zfcp_qdio_send() will unconditionally reduce the timeout to 1 HZ, rather
than leaving it at 10 Hz if it was last armed by the tasklet. This just
makes things more consistent. Also note that we can drop a lot of the
accumulated cruft in qdio_outbound_tasklet(), as zfcp doesn't even use PCI
interrupt requests any longer.
This also slightly touches the Response Queue processing, as
qdio_get_next_buffers() will no longer implicitly scan for Request Queue
completions. So complete the migration to qdio_inspect_queue() here as well
and make the tasklet_schedule() visible.
Link: https://lore.kernel.org/r/018d3ddd029f8d6ac00cf4184880288c637c4fd1.1618417667.git.bblock@linux.ibm.com
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
Place the put_device() call after device_unregister() in both
zfcp_unit_remove() and zfcp_sysfs_port_remove_store() to make it more
natural. put_device() ought to be the last time we touch the object in both
functions.
Add comments after put_device() to make code clearer.
Link: https://lore.kernel.org/r/0a568c7733ba0f1dde28b0c663b90270d44dd540.1618417667.git.bblock@linux.ibm.com
Suggested-by: Steffen Maier <maier@linux.ibm.com>
Suggested-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Qinglang Miao <miaoqinglang@huawei.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
The error path from zfcp_adapter_enqueue() no longer attempts to remove the
diagnostics attributes if they haven't been created yet.
So remove the manual 'sysfs_established' guard for this case, and use
device_add_groups() to add all adapter-related sysfs attributes in one go.
Link: https://lore.kernel.org/r/37a97537f675d643006271f37723c346189b6eec.1618417667.git.bblock@linux.ibm.com
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
When zfcp_adapter_enqueue() fails to create the zfcp_sysfs_adapter_attrs
group, it calls zfcp_adapter_unregister() to tear down the adapter state
again. This then unconditionally attempts to remove the
zfcp_sysfs_adapter_attrs group, resulting in a "group not found" WARN from
sysfs code.
Avoid this by copying most of zfcp_adapter_unregister() into the error
path, allowing for more fine-granular roll-back. Then skip the sysfs
tear-down steps if we haven't progressed this far in the initialization.
Link: https://lore.kernel.org/r/790922cc3af075795fff9a4b787e6bda19bdb3be.1618417667.git.bblock@linux.ibm.com
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
Code indentation should use tabs where possible.
Link: https://lore.kernel.org/r/e8a15a2f3d64e2e76a214647cfd4fe23d370b165.1618417667.git.bblock@linux.ibm.com
Signed-off-by: Yevhen Viktorov <yevhen.viktorov@virginmedia.com>
Signed-off-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
INIT_LIST_HEAD() is only needed for actual list heads, while req->list is
used as a list entry.
Note that when the error path in zfcp_fsf_req_send() removes the request
from the adapter's list of pending requests, it actually looks up the
request from the zfcp_reqlist - rather than just calling list_del(). So
there's no risk of us calling list_del() on a request that hasn't been
added to any list yet.
Link: https://lore.kernel.org/r/254dc0ae28dccc43ab0b1079ef2c8dcb5fe1d2e4.1618417667.git.bblock@linux.ibm.com
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
We are spending way too much effort on qdio-internal bookkeeping for
QAOB management & caching, and it's still not robust. Once qdio's
TX path has detached the QAOB from a PENDING buffer, we lost all
track of it until it shows up in a CQ notification again. So if the
device is torn down before that notification arrives, we leak the QAOB.
Just have the driver take care of it, and simply pass down a QAOB if
they want a TX with async-completion capability. For a buffer in PENDING
state that requires the QAOB for final completion, qeth can now also try
to recycle the buffer's QAOB rather than unconditionally freeing it.
This also eliminates the qdio_outbuf_state array, which was only needed
to transfer the aob->user1 tag from the driver to the qdio layer.
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Acked-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
|
|
COMMAND_COMPLETE is defined as '0', and it is a SCSI parallel message to
boot. So drop the call to set_msg_byte().
Link: https://lore.kernel.org/r/20210113090500.129644-15-hare@suse.de
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
Pull SCSI updates from James Bottomley:
"This consists of the usual driver updates (ufs, qla2xxx, smartpqi,
target, zfcp, fnic, mpt3sas, ibmvfc) plus a load of cleanups, a major
power management rework and a load of assorted minor updates.
There are a few core updates (formatting fixes being the big one) but
nothing major this cycle"
* tag 'scsi-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi: (279 commits)
scsi: mpt3sas: Update driver version to 36.100.00.00
scsi: mpt3sas: Handle trigger page after firmware update
scsi: mpt3sas: Add persistent MPI trigger page
scsi: mpt3sas: Add persistent SCSI sense trigger page
scsi: mpt3sas: Add persistent Event trigger page
scsi: mpt3sas: Add persistent Master trigger page
scsi: mpt3sas: Add persistent trigger pages support
scsi: mpt3sas: Sync time periodically between driver and firmware
scsi: qla2xxx: Update version to 10.02.00.104-k
scsi: qla2xxx: Fix device loss on 4G and older HBAs
scsi: qla2xxx: If fcport is undergoing deletion complete I/O with retry
scsi: qla2xxx: Fix the call trace for flush workqueue
scsi: qla2xxx: Fix flash update in 28XX adapters on big endian machines
scsi: qla2xxx: Handle aborts correctly for port undergoing deletion
scsi: qla2xxx: Fix N2N and NVMe connect retry failure
scsi: qla2xxx: Fix FW initialization error on big endian machines
scsi: qla2xxx: Fix crash during driver load on big endian machines
scsi: qla2xxx: Fix compilation issue in PPC systems
scsi: qla2xxx: Don't check for fw_started while posting NVMe command
scsi: qla2xxx: Tear down session if FW say it is down
...
|
|
Pull block updates from Jens Axboe:
"Another series of killing more code than what is being added, again
thanks to Christoph's relentless cleanups and tech debt tackling.
This contains:
- blk-iocost improvements (Baolin Wang)
- part0 iostat fix (Jeffle Xu)
- Disable iopoll for split bios (Jeffle Xu)
- block tracepoint cleanups (Christoph Hellwig)
- Merging of struct block_device and hd_struct (Christoph Hellwig)
- Rework/cleanup of how block device sizes are updated (Christoph
Hellwig)
- Simplification of gendisk lookup and removal of block device
aliasing (Christoph Hellwig)
- Block device ioctl cleanups (Christoph Hellwig)
- Removal of bdget()/blkdev_get() as exported API (Christoph Hellwig)
- Disk change rework, avoid ->revalidate_disk() (Christoph Hellwig)
- sbitmap improvements (Pavel Begunkov)
- Hybrid polling fix (Pavel Begunkov)
- bvec iteration improvements (Pavel Begunkov)
- Zone revalidation fixes (Damien Le Moal)
- blk-throttle limit fix (Yu Kuai)
- Various little fixes"
* tag 'for-5.11/block-2020-12-14' of git://git.kernel.dk/linux-block: (126 commits)
blk-mq: fix msec comment from micro to milli seconds
blk-mq: update arg in comment of blk_mq_map_queue
blk-mq: add helper allocating tagset->tags
Revert "block: Fix a lockdep complaint triggered by request queue flushing"
nvme-loop: use blk_mq_hctx_set_fq_lock_class to set loop's lock class
blk-mq: add new API of blk_mq_hctx_set_fq_lock_class
block: disable iopoll for split bio
block: Improve blk_revalidate_disk_zones() checks
sbitmap: simplify wrap check
sbitmap: replace CAS with atomic and
sbitmap: remove swap_lock
sbitmap: optimise sbitmap_deferred_clear()
blk-mq: skip hybrid polling if iopoll doesn't spin
blk-iocost: Factor out the base vrate change into a separate function
blk-iocost: Factor out the active iocgs' state check into a separate function
blk-iocost: Move the usage ratio calculation to the correct place
blk-iocost: Remove unnecessary advance declaration
blk-iocost: Fix some typos in comments
blktrace: fix up a kerneldoc comment
block: remove the request_queue to argument request based tracepoints
...
|
|
The request_queue can trivially be derived from the request.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Damien Le Moal <damien.lemoal@wdc.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
As part of removing the power management support from
s390 arch, remove PM callbacks from the scsi/zfcp driver.
Signed-off-by: Vineeth Vijayan <vneethv@linux.ibm.com>
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
|
|
As recovery for a lost Version Change event, trigger an Exchange Config
Data cmd to retrieve the current FW version.
Doing so requires process context (as eg. zfcp_qdio_sbal_get() might need
to sleep), so defer from tasklet context into a work item.
Link: https://lore.kernel.org/r/297c7be2944c3714863fcd22d531d910312d29f0.1603908167.git.bblock@linux.ibm.com
Suggested-by: Steffen Maier <maier@linux.ibm.com>
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
Handle notifications for a concurrent change of the FCP Channel firmware.
Update the relevant user-visible fields to provide accurate data.
Link: https://lore.kernel.org/r/d2c7bc57c6cf1b65eabbf7a5d0e3927b9f65647f.1603908167.git.bblock@linux.ibm.com
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
Explain why the plain spin_lock() suffices in current code, even when the
stat_lock is also used by zfcp_qdio_int_req() in tasklet context.
We could also promote the spin_lock() to a spin_lock_irqsave(), but that
would just obfuscate the locking even further.
Link: https://lore.kernel.org/r/b023b1472630f4bf9fce580577d7bb49de89ccbf.1603908167.git.bblock@linux.ibm.com
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
drivers/s390/scsi/zfcp_ext.h: zfcp_sg_free_table - only declaration left
after commit 58f3ead54752 ("scsi: zfcp: move SG table helper from aux to fc
and make them static")
drivers/s390/scsi/zfcp_ext.h: zfcp_sg_setup_table - only declaration left
after commit 58f3ead54752 ("scsi: zfcp: move SG table helper from aux to fc
and make them static")
Link: https://lore.kernel.org/r/6854ae03c5c65805f746774eeb0f2869fcd6c397.1603908167.git.bblock@linux.ibm.com
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
Shift the IRQ tasklet processing from the qdio layer into zfcp. This will
allow for a good amount of cleanups in qdio, and provides future
opportunity to improve the IRQ processing inside zfcp.
We continue to use the qdio layer's internal tasklet/timer mechanism
(ie. scan_threshold etc) to check for Request Queue completions. Initially
we planned to check for such completions after inspecting the Response
Queue - this should typically work, but there's a theoretical race where
the device only presents the Request Queue completions _after_ all Response
Queue processing has finished. If the Request Queue is then also
_completely_ full, we could send no further IOs and thus get no interrupt
that would trigger an inspection of the Request Queue. So for now stick to
the old model, where we can trust that such a race would be recovered by
qdio's internal timer.
Code-flow wise, this establishes two levels of control:
1. The qdio layer will only deliver IRQs to the device driver if the
QDIO_IRQ_DISABLED flag is cleared. zfcp manages this through
qdio_start_irq() / qdio_stop_irq(). The initial state is DISABLED, and
zfcp_qdio_open() schedules zfcp's IRQ tasklet once during startup to
explicitly enable IRQ delivery.
2. The zfcp tasklet is initialized with tasklet_disable(), and only gets
enabled once we open the qdio device. When closing the qdio device, we
must disable the tasklet _before_ disabling IRQ delivery (otherwise a
concurrently running tasklet could re-enable IRQ delivery after we
disabled it).
A final tasklet_kill() during teardown ensures that no lingering
tasklet_schedule() is still accessing the tasklet structure.
Link: https://lore.kernel.org/r/94a765211c48b74a7b91c5e60b158de01db98d43.1603908167.git.bblock@linux.ibm.com
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux
Pull s390 updates from Vasily Gorbik:
- Remove address space overrides using set_fs()
- Convert to generic vDSO
- Convert to generic page table dumper
- Add ARCH_HAS_DEBUG_WX support
- Add leap seconds handling support
- Add NVMe firmware-assisted kernel dump support
- Extend NVMe boot support with memory clearing control and addition of
kernel parameters
- AP bus and zcrypt api code rework. Add adapter configure/deconfigure
interface. Extend debug features. Add failure injection support
- Add ECC secure private keys support
- Add KASan support for running protected virtualization host with
4-level paging
- Utilize destroy page ultravisor call to speed up secure guests
shutdown
- Implement ioremap_wc() and ioremap_prot() with MIO in PCI code
- Various checksum improvements
- Other small various fixes and improvements all over the code
* tag 's390-5.10-1' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux: (85 commits)
s390/uaccess: fix indentation
s390/uaccess: add default cases for __put_user_fn()/__get_user_fn()
s390/zcrypt: fix wrong format specifications
s390/kprobes: move insn_page to text segment
s390/sie: fix typo in SIGP code description
s390/lib: fix kernel doc for memcmp()
s390/zcrypt: Introduce Failure Injection feature
s390/zcrypt: move ap_msg param one level up the call chain
s390/ap/zcrypt: revisit ap and zcrypt error handling
s390/ap: Support AP card SCLP config and deconfig operations
s390/sclp: Add support for SCLP AP adapter config/deconfig
s390/ap: add card/queue deconfig state
s390/ap: add error response code field for ap queue devices
s390/ap: split ap queue state machine state from device state
s390/zcrypt: New config switch CONFIG_ZCRYPT_DEBUG
s390/zcrypt: introduce msg tracking in zcrypt functions
s390/startup: correct early pgm check info formatting
s390: remove orphaned extern variables declarations
s390/kasan: make sure int handler always run with DAT on
s390/ipl: add support to control memory clearing for nvme re-IPL
...
|
|
While reviewing commit 936e6b85da04 ("scsi: zfcp: Fix panic on ERP timeout
for previously dismissed ERP action"), I stumbled over
zfcp_fsf_req_complete() and wondered whether it has similar issues wrt
concurrent modification of req->erp_action by
zfcp_erp_strategy_check_fsfreq().
But a closer look shows that both its two callers [zfcp_fsf_reqid_check(),
zfcp_fsf_req_dismiss_all()] remove the request from the adapter's req_list
under the req_list's lock. Hence we can trust that if
zfcp_erp_strategy_check_fsfreq() concurrently looks up the corresponding
req_id, it won't find this request and is thus unable to modify it while
it's being processed by zfcp_fsf_req_complete().
Add a code comment that hopefully makes this easier for future readers, and
condense the two accesses to ->erp_action that made me trip over this code
path in the first place.
Link: https://lore.kernel.org/r/c500eac301fcbba5af942bbd200f2d6b14e46994.1599765652.git.bblock@linux.ibm.com
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
Use the right helper to avoid poking around in the list's internals.
Link: https://lore.kernel.org/r/ed669555c73aab95b29444c10066f492c0c43391.1599765652.git.bblock@linux.ibm.com
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
Passing a custom name from the device driver is nice - but in practice
it's only zfcp who has been using this. So we might as well hard-code
a naming scheme in the qdio layer, so that qeth also benefits from it.
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
|
|
Before v4.15 commit 75492a51568b ("s390/scsi: Convert timers to use
timer_setup()"), we intentionally only passed zfcp_adapter as context
argument to zfcp_fsf_request_timeout_handler(). Since we only trigger
adapter recovery, it was unnecessary to sync against races between timeout
and (late) completion. Likewise, we only passed zfcp_erp_action as context
argument to zfcp_erp_timeout_handler(). Since we only wakeup an ERP action,
it was unnecessary to sync against races between timeout and (late)
completion.
Meanwhile the timeout handlers get timer_list as context argument and do a
timer-specific container-of to zfcp_fsf_req which can have been freed.
Fix it by making sure that any request timeout handlers, that might just
have started before del_timer(), are completed by using del_timer_sync()
instead. This ensures the request free happens afterwards.
Space time diagram of potential use-after-free:
Basic idea is to have 2 or more pending requests whose timeouts run out at
almost the same time.
req 1 timeout ERP thread req 2 timeout
---------------- ---------------- ---------------------------------------
zfcp_fsf_request_timeout_handler
fsf_req = from_timer(fsf_req, t, timer)
adapter = fsf_req->adapter
zfcp_qdio_siosl(adapter)
zfcp_erp_adapter_reopen(adapter,...)
zfcp_erp_strategy
...
zfcp_fsf_req_dismiss_all
list_for_each_entry_safe
zfcp_fsf_req_complete 1
del_timer 1
zfcp_fsf_req_free 1
zfcp_fsf_req_complete 2
zfcp_fsf_request_timeout_handler
del_timer 2
fsf_req = from_timer(fsf_req, t, timer)
zfcp_fsf_req_free 2
adapter = fsf_req->adapter
^^^^^^^ already freed
Link: https://lore.kernel.org/r/20200813152856.50088-1-maier@linux.ibm.com
Fixes: 75492a51568b ("s390/scsi: Convert timers to use timer_setup()")
Cc: <stable@vger.kernel.org> #4.15+
Suggested-by: Julian Wiedmann <jwi@linux.ibm.com>
Reviewed-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
Pull SCSI updates from James Bottomley:
"This consists of the usual driver updates (ufs, qla2xxx, tcmu, lpfc,
hpsa, zfcp, scsi_debug) and minor bug fixes.
We also have a huge docbook fix update like most other subsystems and
no major update to the core (the few non trivial updates are either
minor fixes or removing an unused feature [scsi_sdb_cache])"
* tag 'scsi-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi: (307 commits)
scsi: scsi_transport_srp: Sanitize scsi_target_block/unblock sequences
scsi: ufs-mediatek: Apply DELAY_AFTER_LPM quirk to Micron devices
scsi: ufs: Introduce device quirk "DELAY_AFTER_LPM"
scsi: virtio-scsi: Correctly handle the case where all LUNs are unplugged
scsi: scsi_debug: Implement tur_ms_to_ready parameter
scsi: scsi_debug: Fix request sense
scsi: lpfc: Fix typo in comment for ULP
scsi: ufs-mediatek: Prevent LPM operation on undeclared VCC
scsi: iscsi: Do not put host in iscsi_set_flashnode_param()
scsi: hpsa: Correct ctrl queue depth
scsi: target: tcmu: Make TMR notification optional
scsi: target: tcmu: Implement tmr_notify callback
scsi: target: tcmu: Fix and simplify timeout handling
scsi: target: tcmu: Factor out new helper ring_insert_padding
scsi: target: tcmu: Do not queue aborted commands
scsi: target: tcmu: Use priv pointer in se_cmd
scsi: target: Add tmr_notify backend function
scsi: target: Modify core_tmr_abort_task()
scsi: target: iscsi: Fix inconsistent debug message
scsi: target: iscsi: Fix login error when receiving
...
|
|
zfcp_qdio_send() and zfcp_qdio_int_req() run concurrently, adding and
completing SBALs on the Request Queue. There's a theoretical race where
zfcp_qdio_int_req() completes a number of SBALs & increments the queue's
free-level _before_ zfcp_qdio_send() was able to decrement it.
This can cause ->req_q_free to momentarily hold a value larger than
QDIO_MAX_BUFFERS_PER_Q. Luckily zfcp_qdio_send() is always called under
->req_q_lock, and all readers of the free-level also take this lock. So we
can trust that zfcp_qdio_send() will clean up such a temporary overflow
before anyone can actually observe it.
But it's still confusing and annoying to worry about. So adjust the code to
avoid this race.
Link: https://lore.kernel.org/r/7f61f59a1f8db270312e64644f9173b8f1ac895f.1593780621.git.bblock@linux.ibm.com
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
Instead of manually moving each element of the unit and port lists into our
temporary on-stack lists, splice them over in one go.
Link: https://lore.kernel.org/r/cacb179f49ece50fd4dce119c61252d632cdc1d4.1593780621.git.bblock@linux.ibm.com
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
We already maintain a pointer to act->adapter. Use it consistently to avoid
any confusion about whose ->erp_ready_head and ->erp_ready_wq we are
accessing.
Link: https://lore.kernel.org/r/d1bb04322f240dee32f4c4a551bc93bc736f4b01.1593780621.git.bblock@linux.ibm.com
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
zfcp no longer uses the qdio PCI flag, update the comment.
Link: https://lore.kernel.org/r/6717c26fc986bff8776d110e27c199b523684c63.1593780621.git.bblock@linux.ibm.com
Fixes: 21ddaa53f92d ("[SCSI] zfcp: Remove PCI flag")
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Reviewed-by: Fedor Loshakov <loshakov@linux.ibm.com>
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
We don't need crypto-grade random numbers for randomized backoffs. Instead
use prandom_u32_max(ep_ro) which generates a pseudo-random number uniformly
distributed in the interval [0, ep_ro).
Link: https://lore.kernel.org/r/8fc7c4c4069ff1783f4a9ccd84a923f581a09ec5.1593780621.git.bblock@linux.ibm.com
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: George Spelvin <lkml@sdf.org>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
Suppose that, for unrelated reasons, FSF requests on behalf of recovery are
very slow and can run into the ERP timeout.
In the case at hand, we did adapter recovery to a large degree. However
due to the slowness a LUN open is pending so the corresponding fc_rport
remains blocked. After fast_io_fail_tmo we trigger close physical port
recovery for the port under which the LUN should have been opened. The new
higher order port recovery dismisses the pending LUN open ERP action and
dismisses the pending LUN open FSF request. Such dismissal decouples the
ERP action from the pending corresponding FSF request by setting
zfcp_fsf_req->erp_action to NULL (among other things)
[zfcp_erp_strategy_check_fsfreq()].
If now the ERP timeout for the pending open LUN request runs out, we must
not use zfcp_fsf_req->erp_action in the ERP timeout handler. This is a
problem since v4.15 commit 75492a51568b ("s390/scsi: Convert timers to use
timer_setup()"). Before that we intentionally only passed zfcp_erp_action
as context argument to zfcp_erp_timeout_handler().
Note: The lifetime of the corresponding zfcp_fsf_req object continues until
a (late) response or an (unrelated) adapter recovery.
Just like the regular response path ignores dismissed requests
[zfcp_fsf_req_complete() => zfcp_fsf_protstatus_eval() => return early] the
ERP timeout handler now needs to ignore dismissed requests. So simply
return early in the ERP timeout handler if the FSF request is marked as
dismissed in its status flags. To protect against the race where
zfcp_erp_strategy_check_fsfreq() dismisses and sets
zfcp_fsf_req->erp_action to NULL after our previous status flag check,
return early if zfcp_fsf_req->erp_action is NULL. After all, the former
ERP action does not need to be woken up as that was already done as part of
the dismissal above [zfcp_erp_action_dismiss()].
This fixes the following panic due to kernel page fault in IRQ context:
Unable to handle kernel pointer dereference in virtual kernel address space
Failing address: 0000000000000000 TEID: 0000000000000483
Fault in home space mode while using kernel ASCE.
AS:000009859238c00b R2:00000e3e7ffd000b R3:00000e3e7ffcc007 S:00000e3e7ffd7000 P:000000000000013d
Oops: 0004 ilc:2 [#1] SMP
Modules linked in: ...
CPU: 82 PID: 311273 Comm: stress Kdump: loaded Tainted: G E X ...
Hardware name: IBM 8561 T01 701 (LPAR)
Krnl PSW : 0404c00180000000 001fffff80549be0 (zfcp_erp_notify+0x40/0xc0 [zfcp])
R:0 T:1 IO:0 EX:0 Key:0 M:1 W:0 P:0 AS:3 CC:0 PM:0 RI:0 EA:3
Krnl GPRS: 0000000000000080 00000e3d00000000 00000000000000f0 0000000000030000
000000010028e700 000000000400a39c 000000010028e700 00000e3e7cf87e02
0000000010000000 0700098591cb67f0 0000000000000000 0000000000000000
0000033840e9a000 0000000000000000 001fffe008d6bc18 001fffe008d6bbc8
Krnl Code: 001fffff80549bd4: a7180000 lhi %r1,0
001fffff80549bd8: 4120a0f0 la %r2,240(%r10)
#001fffff80549bdc: a53e0003 llilh %r3,3
>001fffff80549be0: ba132000 cs %r1,%r3,0(%r2)
001fffff80549be4: a7740037 brc 7,1fffff80549c52
001fffff80549be8: e320b0180004 lg %r2,24(%r11)
001fffff80549bee: e31020e00004 lg %r1,224(%r2)
001fffff80549bf4: 412020e0 la %r2,224(%r2)
Call Trace:
[<001fffff80549be0>] zfcp_erp_notify+0x40/0xc0 [zfcp]
[<00000985915e26f0>] call_timer_fn+0x38/0x190
[<00000985915e2944>] expire_timers+0xfc/0x190
[<00000985915e2ac4>] run_timer_softirq+0xec/0x218
[<0000098591ca7c4c>] __do_softirq+0x144/0x398
[<00000985915110aa>] do_softirq_own_stack+0x72/0x88
[<0000098591551b58>] irq_exit+0xb0/0xb8
[<0000098591510c6a>] do_IRQ+0x82/0xb0
[<0000098591ca7140>] ext_int_handler+0x128/0x12c
[<0000098591722d98>] clear_subpage.constprop.13+0x38/0x60
([<000009859172ae4c>] clear_huge_page+0xec/0x250)
[<000009859177e7a2>] do_huge_pmd_anonymous_page+0x32a/0x768
[<000009859172a712>] __handle_mm_fault+0x88a/0x900
[<000009859172a860>] handle_mm_fault+0xd8/0x1b0
[<0000098591529ef6>] do_dat_exception+0x136/0x3e8
[<0000098591ca6d34>] pgm_check_handler+0x1c8/0x220
Last Breaking-Event-Address:
[<001fffff80549c88>] zfcp_erp_timeout_handler+0x10/0x18 [zfcp]
Kernel panic - not syncing: Fatal exception in interrupt
Link: https://lore.kernel.org/r/20200623140242.98864-1-maier@linux.ibm.com
Fixes: 75492a51568b ("s390/scsi: Convert timers to use timer_setup()")
Cc: <stable@vger.kernel.org> #4.15+
Reviewed-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
At the moment we allocate and register the Scsi_Host object corresponding
to a zfcp adapter (FCP device) very early in the life cycle of the adapter
- even before we fully discover and initialize the underlying
firmware/hardware. This had the advantage that we could already use the
Scsi_Host object, and fill in all its information during said discover and
initialize.
Due to commit 737eb78e82d5 ("block: Delay default elevator initialization")
(first released in v5.4), we noticed a regression that would prevent us
from using any storage volume if zfcp is configured with support for DIF or
DIX (zfcp.dif=1 || zfcp.dix=1). Doing so would result in an illegal memory
access as soon as the first request is sent with such an configuration. As
example for a crash resulting from this:
scsi host0: scsi_eh_0: sleeping
scsi host0: zfcp
qdio: 0.0.1900 ZFCP on SC 4bd using AI:1 QEBSM:0 PRI:1 TDD:1 SIGA: W AP
scsi 0:0:0:0: scsi scan: INQUIRY pass 1 length 36
Unable to handle kernel pointer dereference in virtual kernel address space
Failing address: 0000000000000000 TEID: 0000000000000483
Fault in home space mode while using kernel ASCE.
AS:0000000035c7c007 R3:00000001effcc007 S:00000001effd1000 P:000000000000003d
Oops: 0004 ilc:3 [#1] PREEMPT SMP DEBUG_PAGEALLOC
Modules linked in: ...
CPU: 1 PID: 783 Comm: kworker/u760:5 Kdump: loaded Not tainted 5.6.0-rc2-bb-next+ #1
Hardware name: ...
Workqueue: scsi_wq_0 fc_scsi_scan_rport [scsi_transport_fc]
Krnl PSW : 0704e00180000000 000003ff801fcdae (scsi_queue_rq+0x436/0x740 [scsi_mod])
R:0 T:1 IO:1 EX:1 Key:0 M:1 W:0 P:0 AS:3 CC:2 PM:0 RI:0 EA:3
Krnl GPRS: 0fffffffffffffff 0000000000000000 0000000187150120 0000000000000000
000003ff80223d20 000000000000018e 000000018adc6400 0000000187711000
000003e0062337e8 00000001ae719000 0000000187711000 0000000187150000
00000001ab808100 0000000187150120 000003ff801fcd74 000003e0062336a0
Krnl Code: 000003ff801fcd9e: e310a35c0012 lt %r1,860(%r10)
000003ff801fcda4: a7840010 brc 8,000003ff801fcdc4
#000003ff801fcda8: e310b2900004 lg %r1,656(%r11)
>000003ff801fcdae: d71710001000 xc 0(24,%r1),0(%r1)
000003ff801fcdb4: e310b2900004 lg %r1,656(%r11)
000003ff801fcdba: 41201018 la %r2,24(%r1)
000003ff801fcdbe: e32010000024 stg %r2,0(%r1)
000003ff801fcdc4: b904002b lgr %r2,%r11
Call Trace:
[<000003ff801fcdae>] scsi_queue_rq+0x436/0x740 [scsi_mod]
([<000003ff801fcd74>] scsi_queue_rq+0x3fc/0x740 [scsi_mod])
[<00000000349c9970>] blk_mq_dispatch_rq_list+0x390/0x680
[<00000000349d1596>] blk_mq_sched_dispatch_requests+0x196/0x1a8
[<00000000349c7a04>] __blk_mq_run_hw_queue+0x144/0x160
[<00000000349c7ab6>] __blk_mq_delay_run_hw_queue+0x96/0x228
[<00000000349c7d5a>] blk_mq_run_hw_queue+0xd2/0xe0
[<00000000349d194a>] blk_mq_sched_insert_request+0x192/0x1d8
[<00000000349c17b8>] blk_execute_rq_nowait+0x80/0x90
[<00000000349c1856>] blk_execute_rq+0x6e/0xb0
[<000003ff801f8ac2>] __scsi_execute+0xe2/0x1f0 [scsi_mod]
[<000003ff801fef98>] scsi_probe_and_add_lun+0x358/0x840 [scsi_mod]
[<000003ff8020001c>] __scsi_scan_target+0xc4/0x228 [scsi_mod]
[<000003ff80200254>] scsi_scan_target+0xd4/0x100 [scsi_mod]
[<000003ff802d8b96>] fc_scsi_scan_rport+0x96/0xc0 [scsi_transport_fc]
[<0000000034245ce8>] process_one_work+0x458/0x7d0
[<00000000342462a2>] worker_thread+0x242/0x448
[<0000000034250994>] kthread+0x15c/0x170
[<0000000034e1979c>] ret_from_fork+0x30/0x38
INFO: lockdep is turned off.
Last Breaking-Event-Address:
[<000003ff801fbc36>] scsi_add_cmd_to_list+0x9e/0xa8 [scsi_mod]
Kernel panic - not syncing: Fatal exception: panic_on_oops
While this issue is exposed by the commit named above, this is only by
accident. The real issue exists for longer already - basically since it's
possible to use blk-mq via scsi-mq, and blk-mq pre-allocates all requests
for a tag-set during initialization of the same. For a given Scsi_Host
object this is done when adding the object to the midlayer
(`scsi_add_host()` and such). In `scsi_mq_setup_tags()` the midlayer
calculates how much memory is required for a single scsi_cmnd, and its
additional data, which also might include space for additional protection
data - depending on whether the Scsi_Host has any form of protection
capabilities (`scsi_host_get_prot()`).
The problem is now thus, because zfcp does this step before we actually
know whether the firmware/hardware has these capabilities, we don't set any
protection capabilities in the Scsi_Host object. And so, no space is
allocated for additional protection data for requests in the Scsi_Host
tag-set.
Once we go through discover and initialize the FCP device firmware/hardware
fully (this is done via the firmware commands "Exchange Config Data" and
"Exchange Port Data") we find out whether it actually supports DIF and DIX,
and we set the corresponding capabilities in the Scsi_Host object (in
`zfcp_scsi_set_prot()`). Now the Scsi_Host potentially has protection
capabilities, but the already allocated requests in the tag-set don't have
any space allocated for that.
When we then trigger target scanning or add scsi_devices manually, the
midlayer will use requests from that tag-set, and before sending most
requests, it will also call `scsi_mq_prep_fn()`. To prepare the scsi_cmnd
this function will check again whether the used Scsi_Host has any
protection capabilities - and now it potentially has - and if so, it will
try to initialize the assumed to be preallocated structures and thus it
causes the crash, like shown above.
Before delaying the default elevator initialization with the commit named
above, we always would also allocate an elevator for any scsi_device before
ever sending any requests - in contrast to now, where we do it after
device-probing. That elevator in turn would have its own tag-set, and that
is initialized after we went through discovery and initialization of the
underlying firmware/hardware. So requests from that tag-set can be
allocated properly, and if used - unless the user changes/disabled the
default elevator - this would hide the underlying issue.
To fix this for any configuration - with or without an elevator - we move
the allocation and registration of the Scsi_Host object for a given FCP
device to after the first complete discovery and initialization of the
underlying firmware/hardware. By doing that we can make all basic
properties of the Scsi_Host known to the midlayer by the time we call
`scsi_add_host()`, including whether we have any protection capabilities.
To do that we have to delay all the accesses that we would have done in the
past during discovery and initialization, and do them instead once we are
finished with it. The previous patches ramp up to this by fencing and
factoring out all these accesses, and make it possible to re-do them later
on. In addition we make also use of the diagnostic buffers we recently
added with
commit 92953c6e0aa7 ("scsi: zfcp: signal incomplete or error for sync exchange config/port data")
commit 7e418833e689 ("scsi: zfcp: diagnostics buffer caching and use for exchange port data")
commit 088210233e6f ("scsi: zfcp: add diagnostics buffer for exchange config data")
(first released in v5.5), because these already cache all the information
we need for that "re-do operation" - the information cached are always
updated during xconf or xport data, so it won't be stale.
In addition to the move and re-do, this patch also updates the
function-documentation of `zfcp_scsi_adapter_register()` and changes how it
reports if a Scsi_Host object already exists. In that case future
recovery-operations can skip this step completely and behave much like they
would do in the past - zfcp does not release a once allocated Scsi_Host
object unless the corresponding FCP device is deconstructed completely.
Link: https://lore.kernel.org/r/030dd6da318bbb529f0b5268ec65cebcd20fc0a3.1588956679.git.bblock@linux.ibm.com
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
When setting an adapter online for the first time, we also create a couple
of entries for it in the sysfs device tree. This is also true even if the
adapter has not yet ever gone successfully through exchange config and
exchange port data.
When moving the scsi host object allocation and registration to after the
first exchange config and exchange port data, this make the `port_rescan`
attribute susceptible to invalid pointer-dereferences of the shost field
before the adapter is fully initialized.
When written to, it schedules a `scan_work` item that will in turn make use
of the associated fibre channel host object to check the topology used for
this FCP device.
Because scanning for remote ports can't be done successfully without
completing exchange config and exchange port data first, we can simply
fence `port_rescan`, and so prevent the illegal access.
As with cases where we can't get a reference to the adapter, we also return
-ENODEV here. Applications need to handle that errno today already.
After a successful allocation of the scsi host object nothing changes in
the work flow.
Link: https://lore.kernel.org/r/ef65366d309993ca91b6917727590ca7ca166c8f.1588956679.git.bblock@linux.ibm.com
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
Common status flags that all main objects - adapter, port, and unit -
support are propagated to sub-objects when set or cleared. For instance,
when setting the status ZFCP_STATUS_COMMON_ERP_INUSE for an adapter object,
we will propagate this to all its child ports and units - same for when
clearing a common status flag.
Units of an adapter object are enumerated via __shost_for_each_device()
over the scsi host object of the corresponding adapter.
Once we move the scsi host object allocation and registration to after the
first exchange config and exchange port data, this won't be possible for
cases where we set or clear common statuses during the very first adapter
recovery.
But since we won't have any port or unit objects yet at that point of time,
we can just fence the status propagation for cases where the scsi host
object is not yet set in the adapter object. It won't change any effective
status propagations, but will prevent us from dereferencing invalid
pointers.
For any later point in the work flow the scsi host object will be set and
thus nothing is changed then.
Link: https://lore.kernel.org/r/f51fe5f236a1e3d1ce53379c308777561bfe35e1.1588956679.git.bblock@linux.ibm.com
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
When doing the very first adapter recovery - initialization - for a FCP
device in a point-to-point topology we also allocate the port object
corresponding to the attached remote port, and trigger a port recovery for
it that will run after the adapter recovery finished.
Right now this happens right after we finished with the exchange config
data command, and uses the fibre channel host object corresponding to the
FCP device to determine whether a point-to-point topology is used.
When moving the scsi host object allocation and registration - and thus
also the fibre channel host object allocation - to after the first exchange
config and exchange port data, this use of the fc_host object is not
possible anymore at that point in the work flow.
But the allocation and recovery trigger doesn't have notable side-effects
on the following exchange port data processing, so we can move those to
after xport data, and thus also to after the scsi host object allocation,
once we move it. Then the fc_host object can be used again, like it is now.
For any further adapter recoveries this doesn't change anything, because at
that point the port object already exists and recovery is triggered
elsewhere for existing port objects.
Link: https://lore.kernel.org/r/73e5d4ac21e2b37bf0c3ca8e530bc5a5c6e74f8f.1588956679.git.bblock@linux.ibm.com
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
When receiving a notification that a FCP device lost its local link we
usually update the fibre channel host object which represents that FCP
device to reflect that.
This notification/information can also surface when the FCP device is
running through adapter recovery (exchange config and exchange port data
return incomplete).
When moving the scsi host object allocation and registration - and thus
also the fibre channel host object allocation - to after the first exchange
config and exchange port data, and this happens during the very first
adapter recovery, these updates can not be done until after the scsi host
object is allocated.
Reorder the fc_host updates in zfcp_fsf_fc_host_link_down() so that they
only happen after a check of whether the scsi host object is already
allocated or not.
During the first adapter recovery this will cause the skip of these updates
if a link-down condition is detected, but we can repeat them after we
allocated the scsi host object, if necessary.
For any further link-down handling the only changes in the work flow are
the slightly reordered assignments in zfcp_fsf_fc_host_link_down().
Link: https://lore.kernel.org/r/f841f2cda61dcd7b8549910c44e1831927459edf.1588956679.git.bblock@linux.ibm.com
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
When executing exchange port data for a FCP device for the first time, or
after an adapter recovery, we update several properties of the fibre
channel host object which represents that FCP device.
When moving the scsi host object allocation and registration - and thus
also the fibre channel host object allocation - to after the first exchange
config and exchange port data, this is not possible for the former case.
Move all these update into separate, and fenced function that first checks
whether the scsi host object already exists or not, before making the
updates.
During the first ever exchange port data in the adapter life cycle this
will make the exchange port data handler skip over this update step, but we
can repeat it later, after we allocated the scsi host object.
For any further recovery of that adapter the work flow is only changed
slightly because then the scsi host object already exists and we don't free
it until we release the adapter completely at the end of its life cycle.
Link: https://lore.kernel.org/r/ae454c2dc6da0b02907c489af91d0b211d331825.1588956679.git.bblock@linux.ibm.com
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
|
When executing exchange config data for a FCP device for the first time, or
after an adapter recovery, we update several properties of the scsi host or
fibre channel host object that represent that FCP device.
When moving the scsi host object allocation and registration - and thus
also the fibre channel host object allocation - to after the first exchange
config and exchange port data, this is not possible for the former case.
Move all these update into separate, and fenced function that first checks
whether the scsi host object already exists or not, before making the
updates.
During the first ever exchange config data in the adapter life cycle this
will make the exchange config data handler skip over this update step, but
we can repeat it later, after we allocated the scsi host object.
For any further recovery of that adapter the work flow is only changed
slightly because then the scsi host object already exists and we don't free
it until we release the adapter completely at the end of its life cycle.
Link: https://lore.kernel.org/r/5fc3f4d38d4334f7aa595497c6f7865fb1102e0f.1588956679.git.bblock@linux.ibm.com
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
|
