[ Upstream commit 764907293e ]
While testing live partition mobility, we have observed occasional crashes
of the Linux partition. What we've seen is that during the live migration,
for specific configurations with large amounts of memory, slow network
links, and workloads that are changing memory a lot, the partition can end
up being suspended for 30 seconds or longer. This resulted in the following
scenario:
CPU 0 CPU 1
------------------------------- ----------------------------------
scsi_queue_rq migration_store
-> blk_mq_start_request -> rtas_ibm_suspend_me
-> blk_add_timer -> on_each_cpu(rtas_percpu_suspend_me
_______________________________________V
|
V
-> IPI from CPU 1
-> rtas_percpu_suspend_me
-> __rtas_suspend_last_cpu
-- Linux partition suspended for > 30 seconds --
-> for_each_online_cpu(cpu)
plpar_hcall_norets(H_PROD
-> scsi_dispatch_cmd
-> scsi_times_out
-> scsi_abort_command
-> queue_delayed_work
-> ibmvfc_queuecommand_lck
-> ibmvfc_send_event
-> ibmvfc_send_crq
- returns H_CLOSED
<- returns SCSI_MLQUEUE_HOST_BUSY
-> __blk_mq_requeue_request
-> scmd_eh_abort_handler
-> scsi_try_to_abort_cmd
- returns SUCCESS
-> scsi_queue_insert
Normally, the SCMD_STATE_COMPLETE bit would protect against the command
completion and the timeout, but that doesn't work here, since we don't
check that at all in the SCSI_MLQUEUE_HOST_BUSY path.
In this case we end up calling scsi_queue_insert on a request that has
already been queued, or possibly even freed, and we crash.
The patch below simply increases the default I/O timeout to avoid this race
condition. This is also the timeout value that nearly all IBM SAN storage
recommends setting as the default value.
Link: https://lore.kernel.org/r/1610463998-19791-1-git-send-email-brking@linux.vnet.ibm.com
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit b2b0f16fa6 ]
A race condition exists between the response handler getting called because
of exchange_mgr_reset() (which clears out all the active XIDs) and the
response we get via an interrupt.
Sequence of events:
rport ba0200: Port timeout, state PLOGI
rport ba0200: Port entered PLOGI state from PLOGI state
xid 1052: Exchange timer armed : 20000 msecs xid timer armed here
rport ba0200: Received LOGO request while in state PLOGI
rport ba0200: Delete port
rport ba0200: work event 3
rport ba0200: lld callback ev 3
bnx2fc: rport_event_hdlr: event = 3, port_id = 0xba0200
bnx2fc: ba0200 - rport not created Yet!!
/* Here we reset any outstanding exchanges before
freeing rport using the exch_mgr_reset() */
xid 1052: Exchange timer canceled
/* Here we got two responses for one xid */
xid 1052: invoking resp(), esb 20000000 state 3
xid 1052: invoking resp(), esb 20000000 state 3
xid 1052: fc_rport_plogi_resp() : ep->resp_active 2
xid 1052: fc_rport_plogi_resp() : ep->resp_active 2
Skip the response if the exchange is already completed.
Link: https://lore.kernel.org/r/20201215194731.2326-1-jhasan@marvell.com
Signed-off-by: Javed Hasan <jhasan@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit e92643db51 ]
If UFS host device is in runtime-suspended state while UFS shutdown
callback is invoked, UFS device shall be resumed for register
accesses. Currently only UFS local runtime resume function will be invoked
to wake up the host. This is not enough because if someone triggers
runtime resume from block layer, then race may happen between shutdown and
runtime resume flow, and finally lead to unlocked register access.
To fix this, in ufshcd_shutdown(), use pm_runtime_get_sync() instead of
resuming UFS device by ufshcd_runtime_resume() "internally" to let runtime
PM framework manage the whole resume flow.
Link: https://lore.kernel.org/r/20201119062916.12931-1-stanley.chu@mediatek.com
Fixes: 57d104c153 ("ufs: add UFS power management support")
Reviewed-by: Can Guo <cang@codeaurora.org>
Signed-off-by: Stanley Chu <stanley.chu@mediatek.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit fe0a8a95e7 ]
iSCSI NOPs are sometimes "lost", mistakenly sent to the user-land iscsid
daemon instead of handled in the kernel, as they should be, resulting in a
message from the daemon like:
iscsid: Got nop in, but kernel supports nop handling.
This can occur because of the new forward- and back-locks, and the fact
that an iSCSI NOP response can occur before processing of the NOP send is
complete. This can result in "conn->ping_task" being NULL in
iscsi_nop_out_rsp(), when the pointer is actually in the process of being
set.
To work around this, we add a new state to the "ping_task" pointer. In
addition to NULL (not assigned) and a pointer (assigned), we add the state
"being set", which is signaled with an INVALID pointer (using "-1").
Link: https://lore.kernel.org/r/20201106193317.16993-1-leeman.duncan@gmail.com
Reviewed-by: Mike Christie <michael.christie@oracle.com>
Signed-off-by: Lee Duncan <lduncan@suse.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 831e3405c2 ]
The current scanning mechanism is supposed to fall back to a synchronous
host scan if an asynchronous scan is in progress. However, this rule isn't
strictly respected, scsi_prep_async_scan() doesn't hold scan_mutex when
checking shost->async_scan. When scsi_scan_host() is called concurrently,
two async scans on same host can be started and a hang in do_scan_async()
is observed.
Fixes this issue by checking & setting shost->async_scan atomically with
shost->scan_mutex.
Link: https://lore.kernel.org/r/20201010032539.426615-1-ming.lei@redhat.com
Cc: Christoph Hellwig <hch@lst.de>
Cc: Ewan D. Milne <emilne@redhat.com>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Lee Duncan <lduncan@suse.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 38b2db564d ]
The be_fill_queue() function can only fail when "eq_vaddress" is NULL and
since it's non-NULL here that means the function call can't fail. But
imagine if it could, then in that situation we would want to store the
"paddr" so that dma memory can be released.
Link: https://lore.kernel.org/r/20200928091300.GD377727@mwanda
Fixes: bfead3b2cb ("[SCSI] be2iscsi: Adding msix and mcc_rings V3")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 4cb9e1ddaa ]
Coverity reported a memory corruption error for the fdmi attributes
routines:
CID 15768 [Memory Corruption] Out-of-bounds access on FDMI
Sloppy coding of the fmdi structures. In both the lpfc_fdmi_attr_def and
lpfc_fdmi_reg_port_list structures, a field was placed at the start of
payload that may have variable content. The field was given an arbitrary
type (uint32_t). The code then uses the field name to derive an address,
which it used in things such as memset and memcpy. The memset sizes or
memcpy lengths were larger than the arbitrary type, thus coverity reported
an error.
Fix by replacing the arbitrary fields with the real field structures
describing the payload.
Link: https://lore.kernel.org/r/20200128002312.16346-8-jsmart2021@gmail.com
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 39c4f1a965 ]
The driver is occasionally seeing the following SLI Port error, requiring
reset and reinit:
Port Status Event: ... error 1=0x52004a01, error 2=0x218
The failure means an RQ timeout. That is, the adapter had received
asynchronous receive frames, ran out of buffer slots to place the frames,
and the driver did not replenish the buffer slots before a timeout
occurred. The driver should not be so slow in replenishing buffers that a
timeout can occur.
When the driver received all the frames of a sequence, it allocates an IOCB
to put the frames in. In a situation where there was no IOCB available for
the frame of a sequence, the RQ buffer corresponding to the first frame of
the sequence was not returned to the FW. Eventually, with enough traffic
encountering the situation, the timeout occurred.
Fix by releasing the buffer back to firmware whenever there is no IOCB for
the first frame.
[mkp: typo]
Link: https://lore.kernel.org/r/20200128002312.16346-2-jsmart2021@gmail.com
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 7b08e89f98 ]
The driver is unable to successfully login with remote device. During pt2pt
login, the driver completes its FLOGI request with the remote device having
WWN precedence. The remote device issues its own (delayed) FLOGI after
accepting the driver's and, upon transmitting the FLOGI, immediately
recognizes it has already processed the driver's FLOGI thus it transitions
to sending a PLOGI before waiting for an ACC to its FLOGI.
In the driver, the FLOGI is received and an ACC sent, followed by the PLOGI
being received and an ACC sent. The issue is that the PLOGI reception
occurs before the response from the adapter from the FLOGI ACC is
received. Processing of the PLOGI sets state flags to perform the REG_RPI
mailbox command and proceed with the rest of discovery on the port. The
same completion routine used by both FLOGI and PLOGI is generic in
nature. One of the things it does is clear flags, and those flags happen to
drive the rest of discovery. So what happened was the PLOGI processing set
the flags, the FLOGI ACC completion cleared them, thus when the PLOGI ACC
completes it doesn't see the flags and stops.
Fix by modifying the generic completion routine to not clear the rest of
discovery flag (NLP_ACC_REGLOGIN) unless the completion is also associated
with performing a mailbox command as part of its handling. For things such
as FLOGI ACC, there isn't a subsequent action to perform with the adapter,
thus there is no mailbox cmd ptr. PLOGI ACC though will perform REG_RPI
upon completion, thus there is a mailbox cmd ptr.
Link: https://lore.kernel.org/r/20200828175332.130300-3-james.smart@broadcom.com
Co-developed-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 53de092f47 ]
It was discovered that sdparm will fail when attempting to disable write
cache on a SATA disk connected via libsas.
In the ATA command set the write cache state is controlled through the SET
FEATURES operation. This is roughly corresponds to MODE SELECT in SCSI and
the latter command is what is used in the SCSI-ATA translation layer. A
subtle difference is that a MODE SELECT carries data whereas SET FEATURES
is defined as a non-data command in ATA.
Set the DMA data direction to DMA_NONE if the requested ATA command is
identified as non-data.
[mkp: commit desc]
Fixes: fa1c1e8f1e ("[SCSI] Add SATA support to libsas")
Link: https://lore.kernel.org/r/1598426666-54544-1-git-send-email-luojiaxing@huawei.com
Reviewed-by: John Garry <john.garry@huawei.com>
Reviewed-by: Jason Yan <yanaijie@huawei.com>
Signed-off-by: Luo Jiaxing <luojiaxing@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit 233bde21aa upstream.
It happens often while I'm preparing a patch for a block driver that
I'm wondering: is a definition of SECTOR_SIZE and/or SECTOR_SHIFT
available for this driver? Do I have to introduce definitions of these
constants before I can use these constants? To avoid this confusion,
move the existing definitions of SECTOR_SIZE and SECTOR_SHIFT into the
<linux/blkdev.h> header file such that these become available for all
block drivers. Make the SECTOR_SIZE definition in the uapi msdos_fs.h
header file conditional to avoid that including that header file after
<linux/blkdev.h> causes the compiler to complain about a SECTOR_SIZE
redefinition.
Note: the SECTOR_SIZE / SECTOR_SHIFT / SECTOR_BITS definitions have
not been removed from uapi header files nor from NAND drivers in
which these constants are used for another purpose than converting
block layer offsets and sizes into a number of sectors.
Cc: David S. Miller <davem@davemloft.net>
Cc: Mike Snitzer <snitzer@redhat.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Nitin Gupta <ngupta@vflare.org>
Reviewed-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 93b6c5db06 ]
In ufshcd_suspend(), after clk-gating is suspended and link is set
as Hibern8 state, ufshcd_hold() is still possibly invoked before
ufshcd_suspend() returns. For example, MediaTek's suspend vops may
issue UIC commands which would call ufshcd_hold() during the command
issuing flow.
Now if UFSHCD_CAP_HIBERN8_WITH_CLK_GATING capability is enabled,
then ufshcd_hold() may enter infinite loops because there is no
clk-ungating work scheduled or pending. In this case, ufshcd_hold()
shall just bypass, and keep the link as Hibern8 state.
Link: https://lore.kernel.org/r/20200809050734.18740-1-stanley.chu@mediatek.com
Reviewed-by: Avri Altman <avri.altman@wdc.com>
Co-developed-by: Andy Teng <andy.teng@mediatek.com>
Signed-off-by: Andy Teng <andy.teng@mediatek.com>
Signed-off-by: Stanley Chu <stanley.chu@mediatek.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 03dbfe0668 ]
When vports are deleted, it is observed that there is memory/kthread
leakage as the vport isn't fully being released.
There is a shost reference taken in scsi_add_host_dma that is not released
during scsi_remove_host. It was noticed that other drivers resolve this by
doing a scsi_host_put after calling scsi_remove_host.
The vport_delete routine is taking two references one that corresponds to
an access to the scsi_host in the vport_delete routine and another that is
released after the adapter mailbox command completes that destroys the VPI
that corresponds to the vport.
Remove one of the references taken such that the second reference that is
put will complete the missing scsi_add_host_dma reference and the shost
will be terminated.
Link: https://lore.kernel.org/r/20200630215001.70793-8-jsmart2021@gmail.com
Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com>
Signed-off-by: James Smart <jsmart2021@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 5aee52c44d ]
clang static analysis flags several null function pointer problems.
drivers/scsi/scsi_transport_spi.c:374:1: warning: Called function pointer is null (null dereference) [core.CallAndMessage]
spi_transport_max_attr(offset, "%d\n");
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Reviewing the store_spi_store_max macro
if (i->f->set_##field)
return -EINVAL;
should be
if (!i->f->set_##field)
return -EINVAL;
Link: https://lore.kernel.org/r/20200627133242.21618-1-trix@redhat.com
Reviewed-by: James Bottomley <jejb@linux.ibm.com>
Signed-off-by: Tom Rix <trix@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 4919b33b63 ]
The adapter info MAD is used to send the client info and receive the host
info as a response. A persistent buffer is used and as such the client info
is overwritten after the response. During the course of a normal adapter
reset the client info is refreshed in the buffer in preparation for sending
the adapter info MAD.
However, in the special case of LPM where we reenable the CRQ instead of a
full CRQ teardown and reset we fail to refresh the client info in the
adapter info buffer. As a result, after Live Partition Migration (LPM) we
erroneously report the host's info as our own.
[mkp: typos]
Link: https://lore.kernel.org/r/20200603203632.18426-1-tyreld@linux.ibm.com
Signed-off-by: Tyrel Datwyler <tyreld@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 7217e6e694 ]
In order to create or activate a new node, lpfc_els_unsol_buffer() invokes
lpfc_nlp_init() or lpfc_enable_node() or lpfc_nlp_get(), all of them will
return a reference of the specified lpfc_nodelist object to "ndlp" with
increased refcnt.
When lpfc_els_unsol_buffer() returns, local variable "ndlp" becomes
invalid, so the refcount should be decreased to keep refcount balanced.
The reference counting issue happens in one exception handling path of
lpfc_els_unsol_buffer(). When "ndlp" in DEV_LOSS, the function forgets to
decrease the refcnt increased by lpfc_nlp_init() or lpfc_enable_node() or
lpfc_nlp_get(), causing a refcnt leak.
Fix this issue by calling lpfc_nlp_put() when "ndlp" in DEV_LOSS.
Link: https://lore.kernel.org/r/1590416184-52592-1-git-send-email-xiyuyang19@fudan.edu.cn
Reviewed-by: Daniel Wagner <dwagner@suse.de>
Reviewed-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Xiyu Yang <xiyuyang19@fudan.edu.cn>
Signed-off-by: Xin Tan <tanxin.ctf@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit e7661a8e5c upstream.
When instrumenting the SCSI layer to run into the
!blk_rq_nr_phys_segments(rq) case the following warning emitted from the
block layer:
blk_peek_request: bad return=-22
This happens because since commit fd3fc0b4d7 ("scsi: don't BUG_ON()
empty DMA transfers") we return the wrong error value from
scsi_prep_fn() back to the block layer.
[mkp: silenced checkpatch]
Signed-off-by: Johannes Thumshirn <jthumshirn@suse.de>
Fixes: fd3fc0b4d7 scsi: don't BUG_ON() empty DMA transfers
Cc: <stable@vger.kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
[iwamatsu: - backport for 4.4.y and 4.9.y
- Use rq->nr_phys_segments instead of blk_rq_nr_phys_segments]
Signed-off-by: Nobuhiro Iwamatsu <nobuhiro1.iwamatsu@toshiba.co.jp>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit b8018b973c upstream.
When checking the model and vendor string we need to use the minimum
value of either string, otherwise we'll miss out on wildcard matches.
And we should take care when matching with zero size strings; results
might be unpredictable. With this patch the rules for matching devinfo
strings are as follows:
- Vendor strings must match exactly
- Empty Model strings will only match if the devinfo model
is also empty
- Model strings shorter than the devinfo model string will
not match
Fixes: 5e7ff2c ("SCSI: fix new bug in scsi_dev_info_list string matching")
Signed-off-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Alan Stern <stern@rowland.harvard.edu>
Reviewed-by: Bart Van Assche <bart.vanassche@wdc.com>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Cc: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
